Fundamentals of Fire Phenomena

Fundamentals of Fire Phenomena Fundamentals of Fire Phenomena James G. Quintiere # 2006 John Wiley & Sons, Ltd ISBN: 0-...

Author: James G. Quintiere

55 downloads 1176 Views 5MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

Fundamentals of Fire Phenomena

Fundamentals of Fire Phenomena James G. Quintiere # 2006 John Wiley & Sons, Ltd ISBN: 0-470-09113-4

Fundamentals of Fire Phenomena

James G. Quintiere University of Maryland, USA

Copyright # 2006

John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex PO19 8SQ, England Telephone (+44) 1243 779777

Email (for orders and customer service enquiries): [email protected] Visit our Home Page on www.wiley.com All Rights Reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning or otherwise, except under the terms of the Copyright, Designs and Patents Act 1988 or under the terms of a licence issued by the Copyright Licensing Agency Ltd, 90 Tottenham Court Road, London W1T 4LP, UK, without the permission in writing of the Publisher. Requests to the Publisher should be addressed to the Permissions Department, John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex PO19 8SQ, England, or emailed to [email protected], or faxed to (+44) 1243 770620. Designations used by companies to distinguish their products are often claimed as trademarks. All brand names and product names used in this book are trade names, service marks, trademarks or registered trademarks of their respective owners. The Publisher is not associated with any product or vendor mentioned in this book. This publication is designed to provide accurate and authoritative information in regard to the subject matter covered. It is sold on the understanding that the Publisher is not engaged in rendering professional services. If professional advice or other expert assistance is required, the services of a competent professional should be sought. Other Wiley Editorial Offices John Wiley & Sons Inc., 111 River Street, Hoboken, NJ 07030, USA Jossey-Bass, 989 Market Street, San Francisco, CA 94103-1741, USA Wiley-VCH Verlag GmbH, Boschstr. 12, D-69469 Weinheim, Germany John Wiley & Sons Australia Ltd, 42 McDougall Street, Milton, Queensland 4064, Australia John Wiley & Sons (Asia) Pte Ltd, 2 Clementi Loop # 02-01, Jin Xing Distripark, Singapore 129809 John Wiley & Sons Canada Ltd, 22 Worcester Road, Etobicoke, Ontario, Canada M9W 1L1 Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic books. Library of Congress Cataloging-in-Publication Data

British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library ISBN-13 978-0-470-09113-5 (HB) ISBN-10 0-470-09113-4 (HB) Typeset in 10/12pt Times by Thomson Press (India) Limited, New Delhi Printed and bound in Great Britain by Antony Rowe Ltd, Chippenham, Wiltshire This book is printed on acid-free paper responsibly manufactured from sustainable forestry in which at least two trees are planted for each one used for paper production.

To those that have brought the logic of science to fire – I have tried to make them part of this book.

Contents

Preface

xiii

Nomenclature

xvii

1

Introduction to Fire 1.1 Fire in History 1.2 Fire and Science 1.3 Fire Safety and Research in the Twentieth Century 1.4 Outlook for the Future 1.5 Introduction to This Book 1.5.1 Thermodynamics 1.5.2 Fluid mechanics 1.5.3 Heat and mass transfer 1.5.4 Supportive references References Problems

1 1 2 8 10 11 13 14 15 16 17 17

2

Thermochemistry 2.1 Introduction 2.2 Chemical Reactions 2.3 Gas Mixture 2.4 Conservation Laws for Systems 2.4.1 Constant pressure reaction 2.4.2 Heat of combustion 2.4.3 Adiabatic flame temperature 2.5 Heat of Formation 2.6 Application of Mass and Energy Conservation in Chemical Reactions 2.7 Combustion Products in Fire References Problems

19 19 20 23 25 27 28 29 30 31 35 41 41

3

49 49 50

Conservation Laws for Control Volumes 3.1 Introduction 3.2 The Reynolds Transport Theorem

viii

CONTENTS

3.3 3.4 3.5 3.6 3.7 Problems

Relationship between a Control Volume and System Volume Conservation of Mass Conservation of Mass for a Reacting Species Conservation of Momentum Conservation of Energy for a Control Volume

53 54 56 59 61 70

4

Premixed Flames 4.1 Introduction 4.2 Reaction Rate 4.3 Autoignition 4.4 Piloted Ignition 4.5 Flame Speed, Su 4.5.1 Measurement techniques 4.5.2 Approximate theory 4.5.3 Fuel lean results 4.5.4 Heat loss effects and extinction 4.6 Quenching Diameter 4.7 Flammability Limits 4.8 Empirical Relationships for the Lower Flammability Limit 4.9 A Quantitative Analysis of Ignition, Propagation and Extinction 4.9.1 Autoignition calculations 4.9.2 Piloted ignition calculations 4.9.3 Flame propagation and extinction calculations 4.9.4 Quenching diameter calculations References Problems

77 77 78 80 85 88 89 90 93 93 95 98 102 105 105 107 107 108 109 110

5

Spontaneous Ignition 5.1 Introduction 5.2 Theory of Spontaneous Ignition 5.3 Experimental Methods 5.4 Time for Spontaneous Ignition References Problems

117 117 119 124 127 130 131

6

Ignition of Liquids 6.1 Introduction 6.2 Flashpoint 6.3 Dynamics of Evaporation 6.4 Clausius–Clapeyron Equation 6.5 Evaporation Rates References Problems

135 135 135 137 141 146 154 154

7

159 159

Ignition of Solids 7.1 Introduction

CONTENTS

ix

7.2

Estimate of Ignition Time Components 7.2.1 Chemical time 7.2.2 Mixing time 7.2.3 Pyrolysis 7.3 Pure Conduction Model for Ignition 7.4 Heat Flux in Fire 7.4.1 Typical heat flux levels 7.4.2 Radiation properties of surfaces in fire 7.4.3 Convective heating in fire 7.4.4 Flame radiation 7.4.5 Heat flux measurements 7.4.6 Heat flux boundary conditions 7.5 Ignition in Thermally Thin Solids 7.5.1 Criterion for thermally thin 7.5.2 Thin theory 7.5.3 Measurements for thin materials 7.6 Ignition of a Thermally Thick Solid 7.6.1 Thick theory 7.6.2 Measurements for thick materials 7.6.3 Autoignition and surface ignition 7.7 Ignition Properties of Common Materials References Problems

161 161 162 163 164 166 166 167 167 169 170 170 171 171 172 174 176 176 180 182 184 188 188

8

Fire 8.1 8.2 8.3 8.4 8.5

Spread on Surfaces and Through Solid Media Introduction Surface Flame Spread – The Thermally Thin Case Transient Effects Surface Flame Spread for a Thermally Thick Solid Experimental Considerations for Solid Surface Spread 8.5.1 Opposed flow 8.5.2 Wind-aided 8.6 Some Fundamental Results for Surface Spread 8.7 Examples of Other Flame Spread Conditions 8.7.1 Orientation effects 8.7.2 Porous media 8.7.3 Liquid flame spread 8.7.4 Fire spread through a dwelling References Problems

191 191 194 198 200 202 202 207 210 213 213 215 216 217 219 220

9

Burning Rate 9.1 Introduction 9.2 Diffusive Burning of Liquid Fuels 9.2.1 Stagnant layer 9.2.2 Stagnant layer solution 9.2.3 Burning rate – an eigenvalue

227 227 233 233 237 241

x

CONTENTS

9.3

Diffusion Flame Variables 9.3.1 Concentrations and mixture fractions 9.3.2 Flame temperature and location 9.4 Convective Burning for Specific Flow Conditions 9.5 Radiation Effects on Burning 9.6 Property Values for Burning Rate Calculations 9.7 Suppression and Extinction of Burning 9.7.1 Chemical and physical factors 9.7.2 Suppression by water and diluents 9.8 The Burning Rate of Complex Materials 9.9 Control Volume Alternative to the Theory of Diffusive Burning 9.9.1 Condensed phase 9.9.2 Gas phase 9.10 General Considerations for Extinction Based on Kinetics 9.10.1 A demonstration of the similarity of extinction in premixed and diffusion flames 9.11 Applications to Extinction for Diffusive Burning References Problems

243 243 246 248 255 259 261 261 262 267 269 271 274 277 279 281 285 286

10 Fire Plumes 10.1 Introduction 10.2 Buoyant Plumes 10.2.1 Governing equations 10.2.2 Plume characteristic scales 10.2.3 Solutions 10.3 Combusting Plumes 10.4 Finite Real Fire Effects 10.4.1 Turbulent axial flame temperatures 10.4.2 Plume temperatures 10.4.3 Entrainment rate 10.4.4 Flame height 10.4.5 Jet flames 10.4.6 Flame heights for other geometries 10.5 Transient Aspects of Fire Plumes 10.5.1 Starting plume 10.5.2 Fireball or thermal References Problems

297 297 302 302 306 308 311 313 313 317 319 322 323 325 326 327 328 332 334

11 Compartment Fires 11.1 Introduction 11.1.1 Scope 11.1.2 Phases of fires in enclosures 11.2 Fluid Dynamics 11.2.1 General flow pattern 11.2.2 Vent flows

339 339 340 340 342 342 343

CONTENTS

xi

11.3 Heat Transfer 11.3.1 Convection 11.3.2 Conduction 11.3.3 Radiation 11.3.4 Overall wall heat transfer 11.3.5 Radiation loss from the vent 11.4 Fuel Behavior 11.4.1 Thermal effects 11.4.2 Ventilation effects 11.4.3 Energy release rate (firepower) 11.5 Zone Modeling and Conservation Equations 11.5.1 Conservation relationships 11.5.2 Dimensionless factors in a solution 11.6 Correlations 11.6.1 Developing fires 11.6.2 Fully developed fires 11.7 Semenov Diagrams, Flashover and Instabilities 11.7.1 Fixed area fire 11.7.2 Second item ignition 11.7.3 Spreading fires References Problems

347 348 348 349 351 351 352 352 353 354 355 356 357 358 358 360 365 366 366 368 369 370

12 Scaling and Dimensionless Groups 12.1 Introduction 12.2 Approaches for Establishing Dimensionless Groups 12.2.1 Buckingham pi method 12.2.2 Partial differential equation (PDE) method 12.2.3 Dimensional analysis 12.3 Dimensionless Groups from the Conservation Equations 12.3.1 Conservation of mass 12.3.2 Conservation of momentum 12.3.3 Energy equation 12.3.4 Heat losses 12.3.5 Mass flows 12.3.6 Liquid droplets 12.3.7 Chemical species 12.3.8 Heat flux and inconsistencies 12.3.9 Summary 12.4 Examples of Specific Correlations 12.4.1 Plume interactions with a ceiling 12.4.2 Smoke filling in a leaky compartment 12.4.3 Burning rate 12.4.4 Compartment fire temperature 12.4.5 Effect of water sprays on fire

377 377 378 379 379 380 380 381 381 382 384 385 386 388 389 392 394 395 396 397 398 400

xii

CONTENTS

12.5 Scale Modeling 12.5.1 Froude modeling 12.5.2 Analog scaling methods References

401 402 403 407

Appendix Flammability Properties Archibald Tewarson

409 409 409

Index

435

Preface

Many significant developments in fire science have emanated from New Jersey. Howard Emmons, the father of US fire research, was born in Morristown and went to school at Stevens Institute of Technology in Jersey City; Dick Magee (flame spread measurements) also taught at Stevens; John deRis (opposed flow flame spread) lived in Englewood; Gerry Faeth (turbulent flames) was raised in Teaneck; Frank Steward (porous media flame spread) and Forman Williams (modeling) hailed from New Brunswick; Bob Altenkirch (microgravity flame spread) is President of New Jersey Institute of Technology (formerly NCE); the FAA has their key fire laboratory for aircraft safety in Atlantic City (Gus Sarkos, Dick Hill and Rich Lyon); Yogesh Jaluria (natural convection in fires) teaches at Rutgers; Glenn Corbett lives in Waldwick and teaches fire science at John Jay College; Jim Milke (fire protection) was raised in Cherry Hill; Irv Glassman (liquid flames spread) taught at Princeton University, inspiring many to enter fire research: Fred Dryer (kinetics), Thor Eklund (aircraft), Carlos Fernandez-Pello (flame spread), Takashi Kashiwagi (material flammability), Tom Ohlemiller (smoldering), Kozo Saito (liquid flame spread), Bill Sirignano (numerical modeling) and more; even Chiang L. Tien (radiation in fire), the former Chancellor of the University of California (Berkeley), graduated from Princeton and almost proceeded to join NYU Mechanical Engineering before being attracted to Berkeley. I was born in Passaic, New Jersey, studied at NCE and graduated with a PhD in Mechanical Engineering from NYU. I guess something in the water made me want to write this book. The motivation and style of this text has been shaped by my education. My training at NYU was focused on the theoretical thermal sciences with no emphasis on combustion. My mentor, W. K. Mueller, fostered careful analysis and a rigor in derivations that is needed to reveal a true understanding of the subject. That understanding is necessary to permit a proper application of the subject. Without that understanding, a subject is only utilized by using formulas or, today, computer programs. I have tried to bring that careful and complete approach to this text. It will emphasize the basics of the subject of fire, and the presentation is not intended to be a comprehensive handbook on the subject. Theory at the level of partial differential equations is avoided in favor of a treatment based on control volumes having uniform properties. Therefore, at most, only ordinary differential equations are formulated. However, a theoretical approach to fire is not sufficient for the solution of problems. Theory cannot allow solutions to phenomena in this complex field, but it does establish a basis for experimental correlations. Indeed, the term fire modeling

xiv

PREFACE

applies more to the formulas that have developed from experimental correlations than from theoretical principles alone. It will be seen, throughout this text, and particularly in Chapter 12 on scaling, that theory strengthens and generalizes experimental correlations. Until computers are powerful enough to solve the basic known equations that apply to fire with appropriate data, we will have to rely on this empirical approach. However, as in other complex fields, correlations refined and built upon with theoretical understanding are invaluable engineering problem-solving tools. I honed my knowledge of fire by having the opportunity to be part of the National Bureau of Standards, NBS (now the National Institute of Standards and Technology, NIST), in an era of discovery for fire that began in the early 1970s. In 1974 the Center for Fire Research (CFR) was established under John Lyons at NBS, joining three programs and establishing a model for a nation’s investment in fire research. That combined program, influenced by the attention to standards from Alex Robertson and Irwin Benjamin, innovations such as oxygen consumption calorimetry (Bill Parker, C. Huggett and V. Babauskas) and others, brought a change in the way fire was looked at. Many had thought that science did not apply; fire was too complex, or rules were essential enough. At that same time, the NSF RANN program for fire research at $2 million was led by Ralph Long who had joined the CFR. Professors Hoyt Hottel and Howard Emmons had long championed the need for fire research in the US and now in the 1970s it was coming to fruition. In addition, the Factory Mutual Research Corporation (FMRC), influenced by Emmons, formed a basic research group, first under John Rockett, nearly by Philip Thomas and mostly carried out by Ray Friedman. The FM fire program became a model of engineering science for fire, bringing such stalwarts as G. Heskestad, J. deRis, A. Tewarson and more to the field. The CFR staff numbered close to 120 at its peak and administered then over $4 million in grants to external programs. The President’s National Commission on Fire Prevention and Control published its report of findings, America Burning (1973), which gave a mandate for fire research and improved technology for the fire service. This time of plenty in the 1970s gave us the foundation and legacy to allow me to write this book. My association at CFR with H. Baum, T. Kashiwagi, M. Harkleroad, B. McCaffrey, H. Nelson, W. Parker, J. Raines, W. Rinkinen and K. Steckler all importantly added to my learning experience. The interaction with distinguished academics in the country from some of the leading institutions, and the pleasure to engage visiting scientists – X. Bodart, M. Curtat, Y. Hasemi, M. Kokkala, T. Tanaka, H. Takeda and K. Saito – all brought something to me. This was the climate at CFR and it was not unique to me; it was the way it was for those who were there. In the beginning, we at CFR reached out to learn and set an appropriate course. Understanding fire was ‘in’, and consumer safety was ‘politically correct’. Interactions with those that knew more were encouraged. John Lyons arranged contact with the Japanese through the auspices of the US–Japan Panel on Natural Resources (UJNR). Visits with the distinguished program at the Fire Research Station (FRS) in the UK were made. The FRS was established after World War II and at its peak in the 1960s under Dennis Lawson had Philip H. Thomas (modeling) and David Rasbash (suppression) leading its two divisions. Their collective works were an example and an essential starting point to us all. Japan had long been involved in the science of fire. One only had to experience the devastation to its people and economy from the 1923 Tokyo earthquake

PREFACE

xv

to have a sensitivity to fire safety. That program gave us S. Yokoi (plumes), K. Kawagoe (vent flows), T. Hirano (flame spread) and many more. The UJNR program of exchange in fire that continued for about the next 25 years became a stimulus for researchers in both countries. The decade of 1975 to about 1985 produced a renaissance in fire science. An intersection of people, government support and international exchange brought fire research to a level of productive understanding and practical use. This book was inspired and influenced by that experience and the collective accomplishments of those that took part. For that experience I am grateful and indebted. I hope my style of presentation will properly represent those that made this development possible. This book is intended as a senior level or graduate text following introductory courses in thermodynamics, fluid mechanics, and heat and mass transfer. Students need general calculus with a working knowledge of elementary ordinary differential equations. I believe the presentation in this text is unique in that it is the first fire text to emphasize combustion aspects and to demonstrate the continuity of the subject matter for fire as a discipline. It builds from chemical thermodynamics and the control volume approach to establish the connection between premixed and diffusion flames in the fire growth sequence. It culminates in the system dynamics of compartment fires without embracing the full details of zone models. Those details can be found in the text Enclosure Fire Dynamics by B. Karlsson and J. G. Quintiere. The current text was influenced by the pioneering work of Dougal Drysdale, An Introduction to Fire Dynamics, as that framed the structure of my course in that subject over the last 20 years. It is intended as a pedagogical exposition designed to give the student the ability to look beneath the engineering formulas. In arriving at the key engineering results, sometimes an extensive equation development is used. Hopefully some students may find this development useful, while others might find it distracting. Also the text is not a comprehensive representation of the literature, but will cite key contributions as illustrative or essential for the course development. I am indebted to many for inspiration and support in the preparation of this text, but the tangible support came from a few. Professor Kristian Hertz was kind enough to offer me sabbatical support for three months in 1999 when I began to formally write the chapters of this text at the Denmark Technical University (DTU). That time also marked the initiation of the MS degree program in fire engineering at the DTU to provide the needed technical infrastructure to support Denmark’s legislation on performance codes for fire safety. Hopefully, this book will help to support educational programs such as that at the DTU, although other engineering disciplines might also benefit by adding a fire safety element to their curriculum. The typing of this text was made tedious by the addition of equation series designed to give the student the process of theoretical developments. For that effort, I am grateful to Stephanie Smith for sacrificing her personal time to type this manuscript. Also, Kate Stewart stepped in at the last minute to add help to the typing, especially with the problems and many illustrations. Finally, the Foundation of the Society of Fire Protection Engineering provided some needed financial support to complete the process. Hopefully this text will benefit their members. I would be remiss if I did not acknowledge the students who suffered through early incomplete versions, and tedious renditions on the blackboard. I thank those students offering corrections and encouragement; and a special appreciation to my graduate students who augmented my thinking and gave me

xvi

PREFACE

illustrative results for the text. Lastly, I am especially indebted to John L. Bryan for believing in me enough to give me an opportunity to teach and to join a special Department in 1990. James G. Quintiere The John L. Bryan Chair in Fire Protection Engineering University of Maryland College Park, Maryland, USA

Nomenclature

A b B Bi c C CS CV d D Da E f F Fr g Gr h H hfg j k l L Le m M n n N Nu p Pr

Area, pre-exponential factor S–Z variable (Chapter 9), plume width Spalding B number, dimensionless plume width Biot number Specific heat Coefficient Control surface Control volume Distance Diffusion coefficient, diameter Damkohler number Energy, activation energy Mixture fraction Radiation view factor, force Froude number Gravitational acceleration Grashof number Heat or mass transfer coefficient, enthalpy per unit mass Enthalpy Heat of vaporization Number of flow surfaces Conductivity Length Heat of gasification Lewis number ¼ Pr/Sc Mass Molecular weight Number of droplets Unit normal vector Number of species in mixture Nusselt number Pressure Prandtl number

xviii

q Q r R Re s S Su Sc Sh t T v V w W x X y yi Y z hc hv " FO;f

i

o

NOMENCLATURE

Heat Fire energy (or heat) Stoichiometric mass oxygen to fuel ratio Universal gas constant Reynolds number Stoichiometric air to fuel mass ratio, entropy per unit mass Surface area Laminar burning speed Schmidt number Sherwood number Time Temperature Velocity Volume Velocity in the z direction Width, work Coordinate Mole fraction, radiation fraction (with designated subscript) Coordinate Mass yield of species i per mass of fuel lost to vaporization Mass fraction Vertical coordinate Thermal diffusivity Coefficient of volumetric expansion Ratio of specific heats, variable defined in Chapters 7 and 9 Distance, Damkohler number (Chapter 5) Difference, distance Heat of combustion Heat of vaporization Emissivity Dimensionless temperature, angle ro ðB þ 1Þ=½B ðro þ 1Þ (Chapter 9) Radiation absorption coefficient Wavelength Viscosity Kinematic viscosity ¼ = Stoichiometric coefficient Dimensionless distance 3.1416. . . Dimensionless group Density Stefan–Boltzman constant¼ 5:67 1011 kW/m2 K4 Dimensionless time cp ðTv T1 Þ=L Dimensionless temperature, equivalence ratio

NOMENCLATURE

Subscripts a, air ad b B BL c d dp e f F g i l L m o O2 , ox p py Q r R s st t T u U v w x 1

Air Adiabatic Burning Balloon Boundary layer Combustion, convection, critical Diffusion Dew point External Flame Fuel Gas Species Liquid Lower limit Mass transfer Initial, vent opening Oxygen Products, pressure, pyrolysis front Pyrolysis Quenching Radiation Reaction Surface Stoichiometric Total Thermal Unburned mixture Upper limit Vaporization Wall, water Coordinate direction Ambient

Other a_ a~ a0 a00 a000 a

Rate of a Molar a a per unit length a per unit area a per unit volume Vector a

xix

1 Introduction to Fire

1.1 Fire in History Fire was recognized from the moment of human consciousness. It was present at the creation of the Universe. It has been a part of us from the beginning. Reliance on fire for warmth, light, cooking and the engine of industry has faded from our daily lives of today, and therefore we have become insensitive to the behavior of fire. Mankind has invested much in the technology to maintain fire, but relatively little to prevent it. Of course, dramatic disastrous fires have been chronicled over recorded history, and they have taught more fear than complete lessons. Probably the low frequency of individually having a bad fire experience has caused us to forget more than we have learned. Fire is in the background of life in the developed world of today compared to being in the forefront of primitive cultures. Fire rarely invades our lives now, so why should we care about it? As society advances, its values depend on what is produced and those sources of production. However, as the means to acquire products becomes easier, values turn inward to the general societal welfare and our environment. Uncontrolled fire can devastate our assets and production sources, and this relates to the societal costs of fire prevention and loss restoration. The effects of fire on people and the environment become social issues that depend on the political ideology and economics that prevail in the state. Thus, attention to fire prevention and control depend on its perceived damage potential and our social values in the state. While these issues have faced all cultures, perhaps the twentieth century ultimately provided the basis for addressing fire with proper science in the midst of significant social and technological advances, especially among the developed countries. In a modern society, the investment in fire safety depends on the informed risk. Reliable risk must be based on complete statistics. An important motivator of the US government’s interest to address the large losses due to fire in the early 1970s was articulated in the report of the National Commission on Fire Prevention and Control (America Burning [1]). It stated that the US annually sustained over $11 billion in lost Fundamentals of Fire Phenomena James G. Quintiere # 2006 John Wiley & Sons, Ltd ISBN: 0-470-09113-4

2

INTRODUCTION TO FIRE

resources and 12 000 lives were lost due to fire. Of these deaths, 3500 were attributed to fires in automobiles; however, a decimal error in the early statistics had these wrong by a factor of 10 (only 350 were due to auto fires). Hence, once funding emerged to address these issues, the true annual death rate of about 8000 was established (a drop in nearly 4000 annual deaths could be attributed to the new research!). The current rate is about 4000 (4126 in 1998 [2]). A big impact on this reduction from about 8000 in 1971 to the current figure is most likely attributed to the increasing use of the smoke detector from nearly none in 1971 to over 66 % after 1981 [3]. The fire death rate reduction appears to correlate with the rate of detector usage. In general, no clear correlation has been established to explain the fire death rates in countries, yet the level of technological and political change seem to be factors. While the estimated cost of fire (property loss, fire department operations, burn injury treatment, insurance cost and productivity loss) was $11.4 billion in the US for 1971 [1], it is currently estimated at 0.875 % of the Gross Domestic Product (GDP) ($11:7 1012 ) or about $102 billion [2]. The current US defense spending is at 3.59 % of GDP. The fire cost per GDP is about the same for most developed countries, ranging from 0.2 to 1.2 % with a mean for 23 countries at 0.81 % [2]. The US is among the highest at per capita fires with 6.5/1000 people and about 1.7 annual deaths per 105 people. Russia tops the latter category among nations at 9.87. This gives a perceived risk of dying by fire in an average lifetime (75 years) at about 1 in 135 for Russia and 1 in 784 for the US. Hence, in the lifetime of a person in the US, about 1 in 800 people will die by fire, about 10 in 800 by auto accidents and the remainder (789) will most likely die due to cancer, heart disease or stroke. These factors affect the way society and governments decide to invest in the wellbeing of its people. Fire, like commercial aircraft disasters, can take a large quantity of life or property cost in one event. When it is realized that 15 to 25 % of fires can be attributed to arson [2], and today terrorism looms high as a threat, the incentive to invest in improved fire safety might increase. The 9/11 events at the World Trade Center (WTC) with a loss of life at nearly 3000 and a direct cost of over $10 billion cannot be overlooked as an arson fire. In the past, such catastrophic events have only triggered ‘quick fixes’ and short-term anxiety. Perhaps, 9/11 – if perceived as a fire safety issue – might produce improved investment in fire safety. Even discounting the WTC as a fire event, significant disasters in the twentieth century have and should impact on our sensitivity to fire. Yet the impact of disasters is short-lived. Events listed in Table 1.1 indicate some significant fire events of the twentieth century. Which do you know about? Related illustrations are shown in Figures 1.1 to 1.4. It is interesting to note that the relatively milder (7.1 versus 7.7) earthquake centered in Loma Prieta in 1989 caused an identical fire in San Francisco’s Marina district, as shown in Figure 1.2(b).

1.2 Fire and Science Over the last 500 years, science has progressed at an accelerating pace from the beginnings of mathematical generality to a full set of conservation principles needed to address most problems. Yet fire, one of the earliest tools of mankind, needed the last 50 years to give it mathematical expression. Fire is indeed complex and that surely helped to retard its scientific development. But first, what is fire? How shall we define it?

3

FIRE AND SCIENCE Table 1.1 Selected fire disasters of the twentieth century Year

Event

1903 1904 1906

Iroquois Theater fire, Chicago, 620 dead Baltimore conflagration, $50 million ($821 million, 1993) San Francisco earthquake (those that were there, call it ‘the fire’), 28 000 buildings lost, 450 dead, $300 million due to fire ($5 billion, 1993), $15 million due to earthquake damage Tokyo earthquake, 150 000 dead and 700 000 homes destroyed mostly by fire, 38 000 killed by a fire tornado (whirl) who sought refuge in park Morro Castle ship fire, 134 dead, Asbury Park, NJ Hindenburg airship fire, Lakehurst, NJ SS Grandcamp and Monsanto Chemical Co. (spontaneous ignition caused fire and explosion), Texas City, TX, $67 million ($430 million, 1993) Apollo One spacecraft, 3 astronauts died, fire in 100 % oxygen Managua, Nicaragua earthquake, 12 000 dead, fire and quake destroyed the entire city MGM Grand Hotel, Las Vegas, NV, 84 dead Australian bushfires, 70 dead, 8500 homeless, 150 000 acres destroyed, $1 billion One Meridian Plaza 38-story high-rise, Philadelphia, 3 fire fighters killed, $500 million loss as structural damage caused full demolition Oakland firestorm, $1.5 billion loss Milliken Carpet Plant, Le Grange, GA, 660 00 ft2 destroyed, full-sprinklers, $500 million loss World Trade Center twin towers, 3000 died, $10 billion loss, insufficient fire protection to steel structure

1923 1934 1937 1947 1967 1972 1980 1983 1991 1991 1995 2001

Source — [5] [4] [4] [4] [4] [5] [4] [4] [4] [4] — [5] — —

A flame is a chemical reaction producing a temperature of the order of at least 1500 K and generally about 2500 K at most in air. Fire is generally a turbulent ensemble of flames (or flamelets). A flamelet or laminar flame can have a thickness of the order of 103 cm and an exothermic production rate of energy per unit volume of about 108 W=cm3 . However, at the onset of ignition, the reaction might only possess

Figure 1.1

1904 Baltimore conflagration

4

INTRODUCTION TO FIRE

Figure 1.2

(a) 1906 San Francisco earthquake. (b) area of fire damage for San Francisco, 1906

about 102 W/cm3. This is hardly perceptible, and its abrupt transition to a full flame represents a jump in thermal conditions, giving rise to the name thermal explosion. A flame could begin with the reactants mixed (premixed) or reactants that might diffuse together (diffusion flame). Generally, a flame is thought of with the reactants in the gas phase. Variations in this viewpoint for a flame or fire process might occur and are defined in special terminology. Indeed, while flame applies to a gas phase reaction, fire,

FIRE AND SCIENCE

Figure 1.3 Oakland Hills fire storm, 1991

Figure 1.4

Collapse of the south tower at the World Trade Center

5

6

INTRODUCTION TO FIRE

and its synonym combustion, refers to a broader class of reactions that constitute a significant energy density rate. For example, smoldering is a combustion reaction (that could occur under temperatures as low as 600 K) between oxygen in air and the surface of a solid fuel. The combustion wave propagation through dynamite might be termed by some as an explosion, yet it is governed by premixed flame theory. Indeed, fire or combustion might more broadly represent an exothermic chemical reaction that results from a runaway rate caused by temperature or catalytic effects. Note that we have avoided the often-used definition of fire as ‘a chemical reaction in air giving off heat and light’. However, a flame may not always be seen; e.g. an H2 flame would be transparent to the eye and not easily seen. A flame could be made adiabatic, and therefore heat is not given off. This could occur within the uniform temperature soot-laden regions of a large fire. Moreover, oxygen in air might not be the only oxidizer in a reaction termed combustion or fire. In general, we might agree that a flame applies to gas phase combustion while fire applies to all aspects of uncontrolled combustion. The science of fire required the development of the mathematical description of the processes that comprise combustion. Let us examine the historical time-line of those necessary developments that began in the 1600s with Isaac Newton. These are listed in Table 1.2, and pertain to macroscopic continuum science (for examples see Figures 1.5 and 1.6). As with problems in convective heat and mass transfer, fire problems did not require profound new scientific discoveries after the general conservation principles and constitutive relations were established. However, fire is among the most complex of transport processes, and did require strategic mathematical formulations to render solutions. It required a thorough knowledge of the underlying processes to isolate its dominant elements in order to describe and effectively interpret experiments and create general mathematical solutions. Table 1.2

Developments leading to fire science

Time

Event

Key initiator

1650 1737 1750 1807 1827 1845 1850

Second law of motion (conservation of momentum) Relationship between pressure and velocity in a fluid First law of thermodynamics (conservation of energy) Heat conduction equation (Fourier’s law) Viscous equations of motion of a fluid

Isaac Newton Daniel Bernoulli Rudolph Clausius Joseph Fourier Navier Stokes Michael Faraday

1855 1884 1900 1928 1940 1930 1950 1950 1960 1970

Chemical History of the Candle Lectures at the Royal Society Mass diffusion equation (Fick’s law) Chemical reaction rate dependence on temperature Thermal radiation heat transfer Solution of diffusion flame in a duct Combustion equations with kinetics Convective burning solutions Fire phenomena solutions Leadership in US fire research programs

A. Fick S. Arrhenius Max Planck Burke and Schumann Frank-Kamenetskii Semenov Zel’dovich H. Emmons, D. B. Spalding P. H. Thomas R. Long, J. Lyons

FIRE AND SCIENCE

7

Figure 1.5 Bernoulli’s treatise on hydrodynamics

Figure 1.6

Significant scientists in (a) heat transfer (Jean Baptiste Joseph Fourier), (b) chemistry (Svante August Arrhenius, Nobel Prize in Chemistry 1903) and (c) combustion (Nikolay Semenov, Nobel Prize in Chemistry 1956)

8

INTRODUCTION TO FIRE

While Newton forged the basis of the momentum equation, he also introduced the concept of mass, essentially defined by m ¼ F=a. Joule established the equivalence between work and energy, and the first law of thermodynamics (energy conservation) introduces the concept of energy as the change due to heat loss and work done by a system. Heat is then defined as that energy transferred from the system due to temperature difference. Of course, the second law of thermodynamics (Clausius) does not enter into solution methods directly, but says a system cannot return to its original state without a cost. That cost involves transfers of mass and energy, and these transport fluxes become (approximately) related to the gradient of their associated intensive properties; e.g. heat flux is proportional to the temperature gradient as in Fourier’s law and species mass flux is proportional to its concentration gradient as in Fick’s law. The nonlinear transport of heat due to radiation provides a big obstacle to obtaining analytical solutions for fire; and turbulence, despite the exact formulation of this unsteady phenomena through the Navier–Stokes equations over 150 years ago, precludes general solutions to fluid flow problems. Hence, skill, ingenuity and broad knowledge are needed to formulate, solve and understand fire phenomena. The Russian scientists Semenov [6], Frank-Kamenetskii [7] and Zel’dovich et al. [8] provide much for the early formulation of combustion problem solutions. These problems addressed the role of kinetics and produced the Zel’dovich number, ðE=RTÞ=ð1 þ cp T=hc Þ, E=RT being the Arrhenius parameter. Spalding [9] and Emmons [10] laid a foundation for solutions to diffusive burning of the condensed phase. This foundation served as a guidepost to other far more complex problems involving soot, radiation and flame spread. The Spalding B number is a key dimensionless group that emerges in these problems, and represents the ratio of energy released in combustion to that needed for fuel vaporization. The energy production rate by fire (more specifically its rate of change of enthalpy due to chemical changes at 25 C and 1 atmosphere pressure) will be termed here as firepower. Many have used the term heat release rate (HRR), but this is viewed as a misnomer by this author as the firepower has the same chemical production rate that would be attributed to a combustion-based power plant, and is not identical to the heat transferred in the reaction. Such misnomers have occurred throughout thermodynamics as represented in the terms: heat capacity, heat of vaporization and heat of combustion – all related to enthalpies. These misnomers might be attributed to remnants of the caloric theory of heat in which heat was viewed as being part of matter.

1.3 Fire Safety and Research in the Twentieth Century At the start of the twentieth century there were significant activities in US fire safety. The National Fire Protection Association (NFPA) was founded in 1897 principally to address sprinkler use and its standardization. The Underwriter’s Laboratory (UL) was founded in 1894 by William H. Merrill to address electrical standards and testing. In 1904, Congress created the National Bureau of Standards (NBS), later becoming the National Institute of Standards and Technology (NIST) in 1987. After the Baltimore conflagration (1904), Congress directed NBS to address the structural fire safety of buildings. That program began in 1914 under Simon Ingberg, and had a profound influence on standards and testing dealing with structural fire protection over the next 60 years. It is noteworthy to consider the words of the NBS Director to Congress in 1914 when US fire losses were

FIRE SAFETY AND RESEARCH IN THE TWENTIETH CENTURY

9

reported at more than ten time those of Europe: ‘The greatest fire loses are in the cities having laws and regulations’ [11]. Sixty years later America Burning cited ‘One basic need is to strengthen the grounding of knowledge about fire in a body of scientific and engineering theory, so that real world problems can be dealt with through predictive analyses’ [1]. The result of that President’s Commission report led to the consolidation of the three existing fire programs at NBS at that time: building fire safety (I. Benjamin and D. Gross), fire research and fire fighting (J. Rockett and A. F. Robertson) and flammable fabrics (J. Clark and C. Huggett). The consolidation forged the Center for Fire Research (CFR) under John Lyons in 1974. It included the development of a library for fire science (M. Rappaport and N. Jason) – a necessity for any new research enterprise. The CFR program focused on basic research and important problems of that time. They involved clothing fabric and furnishing fire standards, the development of smoke detectors standards, floor covering flammability and special standards for housing and health care facilities. Its budget was about $5 million, with more than half allocated as internal NBS funds. It co-opted the NSF RANN (Research for Applied National Needs) fire grants program along with Ralph Long, its originator. This program was endowed solely for basic fire research at $2 million. Shortly after 1975, the Products Research Committee (PRC), a consortium of plastic producers, contributed about $1 million annually to basic research on plastic flammability for the next five years. This industry support was not voluntary, but mandated by an agreement with the Federal Trade Commission that stemmed from a complaint that industry was presenting faulty fire safety claims with respect to test results. This total budget for fire research at NBS of about $8 million in 1975 would be about $30 million in current 2004 dollars. Indeed, the basic grants funding program that originating from the NSF is only about $1 million today at NBS, where inflation would have its originally allocated 1974 amount at about $7.5 million today. Thus, one can see that the extraordinary effort launched for fire research in the 1970s was significant compared to what is done today in the US. While US fire research efforts have waned, international interests still remain high. Fire research in the USA during the 1970s was characterized by a sense of discovery and a desire to interact with centers of knowledge abroad. For years, especially following World War II, the British and the Japanese pursued research on fire with vigor. The Fire Research Station at Borehamwood (the former movie capital of the UK) launched a strong research program in the 1950s with P. H. Thomas, a stalwart of fire modeling [12], D. Rasbash (suppression), M. Law (fire resistance) and D. L. Simms (ignition). The Japanese built a strong infrastructure of fire research within its schools of architecture, academia in general and government laboratories. Kunio Kawagoe was the ambassador of this research for years (about 1960 to 1990). He is known for contributions to compartment fires. Others include S. Yokoi (plumes), T. Hirano (flame spread), T. Wakamatsu (smoke movement), T. Tanaka (zone modeling), Y. Hasemi (fire plumes), T. Jin (smoke visibility), K. Akita (liquids) and H. Takeda (compartment fire behavior). In Sweden, beginning in the early 1970s, Professor O. Pettersson engaged in fire effects on structures, fostering the work of S. E. Magnusson and founding the fire engineering degree program at Lund University. Vihelm Sjolin, coming from the Swedish civil defense program, provided leadership to fire researchers like Kai Odeen (compartment fires) and Bengt Hagglund (radiation, fire dynamics), and later directed a $1 million Swedish basic research program in fire during the 1980s. Today Sweden has strong fire research efforts at Lund University and at the SP Laboratory in Boras (U. Wickstrom and

10

INTRODUCTION TO FIRE

B. Sundstrom). Programs remain active around the world: PR China (W. C. Fan), Finland (M. Kokkala and O. Keski-Rinkinen), France (P. Joulain and P. Vantelon) and the United Kingdom (G. Cox, D. Drysdale, J. Torero, G. Makhviladze, V. Molkov, J. Shields and G. Silcock) to cite only a few. Enabled by ample funding, enthusiasm for problems rich in interdisciplinary elements and challenged by new complex problems, many new researchers engaged in fire research in the 1970s. They were led by Professor Howard Emmons (Harvard), who had been encouraged to enter the field of fire research by Professor Hoyt Hottel (MIT). They both brought stature and credibility to the US fire research interest. Emmons also advised the Factory Mutual Research Corporation (FMRC) on its new basic program in fire research. The FM program inspired by Jim Smith was to serve insurance interests – a significant step for industry. That spawned a remarkably prolific research team: R. Friedman, J. deRis, L. Orloff, G. Markstein, M. Kanury, A. Modak, C. Yao, G. Heskestad, R. Alpert, P. Croce, F. Taminini, M. Delichatsios, A. Tewarson and others. They created a beacon for others to follow. Under NBS and NSF funding, significant academic research efforts emerged in the US at this time. This brought needed intellect and skills from many. It was a pleasure for me to interact with such dedicated and eager scientists. Significant programs emerged at UC Berkeley (C. L. Tien, P. Pagni, R. B. Williamson, A. C. Fernandez-Pello and B. Bressler), Princeton (I. Glassman, F. Dryer, W. Sirignano and F. A. Williams), MIT (T. Y. Toong and G. Williams), Penn State (G. M Faeth), Brown (M. Sibulkin), Case-Western Reserve (J. Tien and J. Prahl) and, of course, Harvard (H. Emmons and G. Carrier). Many students schooled in fire were produced (A. Atreya, I. Wichman, K. Saito, D. Evans and V. Babrauskas – to name only a few). All have made their mark on contributions to fire research. Other programs have contributed greatly: the Naval Research Laboratory (H. Carhart and F. Williams), FAA Technical Center (C. Sarkos, R. Hill, R. Lyon and T. Eklund), SWRI (G. Hartzell and M. Janssens), NRC Canada (T. Harmathy, T. T. Lie and K. Richardson), IITRI (T. Waterman) and SRI (S. Martin and N. Alvares). This collective effort reached its peak during the decade spanning 1975 to 1985. Thereafter, cuts in government spending took their toll, particularly during the Reagan administration, and many programs ended or were forced to abandon their basic research efforts. While programs still remain, they are forced into problem-solving modes to maintain funds for survival. Indeed, the legacy of this fruitful research has been the ability to solve fire problems today.

1.4 Outlook for the Future In 2002 NIST was directed by Congress under $16 million in special funding to investigate the cause of the 9/11 World Trade Center (WTC) building failures. The science of fire, based on the accomplishment of the fertile decade (1975 to 1985), has increasingly been used effectively in fire investigations. The Bureau of Alcohol, Tobacco and Firearms (BATF), having the responsibility of federal arson crimes, has built the largest fire laboratory in the world (2003). The dramatic events of 9/11 might lead to a revitalized public awareness that fire science is an important (missing) element in fire safety regulation tenets. The NIST investigation could show that standards used

INTRODUCTION TO THIS BOOK

11

in regulation for fire safety might be insufficient to the public good. Students of this text will acquire the ability to work problems and find answers, related to aspects of the 1993 bombing and fire at the WTC, and the 9/11 WTC attacks. Some of these problems address issues sometimes misrepresented or missed by media reporting. For example, it can clearly be shown that the jet fuel in the aircraft attacks on the WTC only provided an ignition source of several minutes, yet there are still reports that the jet fuel, endowed with special intensity, was responsible for the collapse of the steel structures. It is generally recognized that the fires at the WTC were primarily due to the normal furnishing content of the floors, and were key in the building failures. It is profound that NIST is now investigating such a significant building failure due to fire as the mandate given by Congress to the NBS founding fire program in 1914 – 90 years ago – that mandate was to improve the structural integrity of buildings in fire and to insure against collapse. While much interest and action have been taken toward the acceptance of performance codes for fire safety in, for example, Japan, Denmark, Australia, New Zealand, the US and the rest of the world, we have not fully created the infrastructure for the proper emergence of performance codes for fire. Of course, significant infrastructure exists for the process of creating codes and standards, but there is no overt agency to insure that fire science is considered in the process. The variations in fire test standards from agency to agency and country to country are a testament to this lack of scientific underpinning in the establishment of regulations. Those that have expertise in fire standards readily know that the standards have little, if any, technical bases. Yet these standards have been generally established by committees under public consensus, albeit with special interests, and their shortcomings are not understood by the general public at large. Even without direct government support or international harmonization strategies, we are evolving toward a better state of fire standards with science. Educational institutions of higher learning probably number over 25 in the world currently. That is a big step since the founding of the BS degree program in fire protection engineering at the University of Maryland in 1956 (the first accredited in the US). This was followed by the first MS program, founded by Dave Rasbash in 1975, at the University of Edinburgh. It is very likely that it will be the force of these programs that eventually lead to the harmonization of international fire standards using science as their underlying principles. This will likely be an uneven process of change, and missteps are likely; to paraphrase Kristian Hertz (DTU, Denmark): ‘It is better to use a weak scientific approach that is transparent enough to build upon for a standard practice, than to adopt a procedure that has no technical foundation, only tradition.’ Hopefully, this text will help the student contribute to these engineering challenges in fire safety standards. Fire problems will continually be present in society, both natural and manmade, and ever changing in a technological world.

1.5 Introduction to This Book This book is intended to be pedagogical, and not inclusive as a total review of the subject. It attempts to demonstrate that the subject of fire is a special engineering discipline built on fundamental principles, classical analyses and unique phenomena. The flow of the

12

INTRODUCTION TO FIRE

subject matter unfolds in an order designed to build upon former material to advance properly to the new material. Except for the first and last chapters, instruction should follow the order of the text. In my teaching experience, a one-semester 15-week, 45-hour course, covering most of Chapters 2 through 11, are possible with omission of specialized or advanced material. However, Chapter 11 on compartment fires is difficult to cover in any depth because the subject is too large, and perhaps warrants a special course that embraces associate computer models. The text by Karlsson and Quintiere, Enclosure Fire Dynamics, is appropriate for such a course (see Section 1.5.4). The current text attempts to address the combustion features of fire engineering in depth. That statement may appear as an oxymoron since fire and combustion are the same, and one might expect them to be fully covered. However, fire protection engineering education has formerly emphasized more the protection aspect than the fire component. Figure 1.7 shows a flow chart that represents how the prerequisite

Thermodynamics

Calculus

Ordinary Differential Equations

F luid Mechanics Heat and Mass Transfer

Chemical Thermodynamics

Control Volumes & Conservation Laws

Premixed Flames

Spontaneous Ignition

Ignition of Liquids and Solids

Flame Spread

Burning Rate for Diffusion Flames

Fire Plumes Compartment Fires

Scaling and Dimensional Analysis

Figure 1.7

Flow chart for subject material

INTRODUCTION TO THIS BOOK

13

material relates to the material in the text. Standard texts can be consulted for the prerequisite subjects. However, it is useful to present a brief review of these subjects to impart the minimum of their facets and to stimulate the student’s recollections.

1.5.1

Thermodynamics

Thermodynamics is the study of matter as a macroscopic continuum. Mass is the tangible content of matter, identified through Newton’s second law of motion, m ¼ F=a. A system of fixed mass which could consist of a collection of different chemical states of matter (species) must remain at a constant total energy if it is isolated from its surroundings. For fire systems only internal energy (U) will be significant (excluding kinetic, electrical, etc.). When a system interacts with its surroundings, the first law of thermodynamics is expressed as U2 U1 ¼ Q þ W

or as a rate expression

dU _ ¼ Q_ þ W dt

ð1:1Þ

The internal energy changes from 1 to 2 and heat (Q) and work (W) are considered positive when added to the system. Work is defined as the product of the resultant of forces and their associated distances moved. Heat is energy transferred solely due to temperature differences. (Sometimes, the energy associated with species diffusion is included in the definition of heat.) The first law expressed in rate form requires the explicit introduction of the transport process due to gradients between the system and the surroundings. Despite these gradients implying nonequilibrium, thermodynamic relationships are assumed to hold at each instant of time, and at each point in the system. The second law of thermodynamics allows us to represent the entropy change of a system as dS ¼

1 p dU þ dV T T

ð1:2Þ

This relationship is expressed in extensive properties that depend on the extent of the system, as opposed to intensive properties that describe conditions at a point in the system. For example, extensive properties are made intensive by expressing them on a per unit mass basis, e.g. s ¼ S=m density, ¼ 1=v, v ¼ V=m. For a pure system (one species), Equation (1.2) in intensive form allows a definition of thermodynamic temperature and pressure in terms of the intensive properties as @u Temperature; T ¼ @s v ð1:3Þ @s Pressure; p ¼ @v u The thermodynamic pressure is equal to the mechanical pressure due to force at equilibrium. While most problems of interest possess gradients in the intensive properties

14

INTRODUCTION TO FIRE

(i.e. T, p, Yi ), local thermodynamic equilibrium is implicitly assumed. Thus, the equation of state, such as for an ideal gas p¼

RT M

ð1:4Þ

holds for thermodynamic properties under equilibrium, although it may be applied over variations in time and space. Here M is the molecular weight, is the density and R is the universal gas constant. For ideal gases and most solids and liquids, the internal energy and enthalpy ðh ¼ u þ p=Þ can be expressed as du ¼ cv dT

and

dh ¼ cp dT

respectively

ð1:5Þ

where cv and cp are the specific heats of constant volume and constant pressure. It can be shown that for an ideal gas, R ¼ cp cv .

1.5.2

Fluid mechanics

Euler expressed Newton’s second law of motion for a frictionless (inviscid) fluid along a streamline as 2 2 @v v v ds þ þ p þ gz þ p þ gz ¼ 0 2 2 1 @t 2 1

ð2

ð1:6Þ

The velocity (v) is tangent along the streamline and z is the vertical height opposite in sign to the gravity force field, g ¼ 9:81 N=kg ð1 N ¼ 1 kg m=s2 Þ. For a viscous fluid, the right-hand side (RHS) is negative and in magnitude represents the power dissipation per unit volume due to viscous effects. The RHS can also be represented as the rate of shaft work added to the streamline or system in general. Pressure is the normal stress for an inviscid fluid. The shear stress for a Newtonian fluid is given as ¼

@v @n

ð1:7Þ

where is the viscosity and n is the normal direction. Shear stress is shown for a streamtube in Figure 1.8. Viscous flow is significant for a small Reynolds number

Figure 1.8

Flow along a streamline

INTRODUCTION TO THIS BOOK

15

ðRe ¼ vl=Þ, where l is a characteristic length. Near surfaces, l might be thought of as the boundary-layer thickness where all of the viscous effects are felt and Re is locally small. However, if l is regarded as a characteristic geometric length, as the chord of an airfoil, Re might be found to be large. This means that the bulk flow away from the surface is nearly inviscid. Flows become unstable at large Re, evolving into a chaotic unsteady flow known as turbulence. Turbulence affects the entire flow domain except very close to the surface, where it is suppressed and laminar flow prevails.

1.5.3

Heat and mass transfer

Heat can flow due to the temperature difference in two distinct mechanisms: (1) radiation and (2) conduction. Radiation is energy transfer by electromagnetic waves and possesses frequency (or wavelength). Radiant energy can be generated in many different ways, which usually identifies its application, such as radio waves, microwaves, gamma rays, X-rays, etc. These different forms are usually identified by their wavelength ranges. Planck’s law gives us the emitted radiation flux due to temperature alone as q_ 00 ¼ " T 4

ð1:8Þ

where " is the emissivity of the radiating body, varying between 0 and 1. The emissivity indicates the departure of an ideal (blackbody) emitter. For solids or liquids under fire heating conditions, although " depends on wavelength, it can generally be regarded as a constant close to 1. For gases, " will be finite only over discrete wavelengths, but for soot distributed in combustion product gas, " is continuous over all wavelengths. It is this soot in flames and combustion products that produces the bulk of the radiation. In general, for smoke and flames the emissivity can be represented for each wavelength as " ¼ 1 e l

ð1:9Þ

The absorption coefficient depends in general on wavelength (). For large l, " approaches 1 – a perfect emitter. This limit generally occurs at turbulent flames of about 1 m in thickness. The other mode of heat transfer is conduction. The conductive heat flux is, by Fourier’s law, @T T

k ð1:10Þ q_ 00 ¼ k @n l For laminar conditions of slow flow, as in candle flames, the heat transfer between a fluid and a surface is predominately conductive. In general, conduction always prevails, but in the unsteadiness of turbulent flow, the time-averaged conductive heat flux between a fluid and a stationary surface is called convection. Convection depends on the flow field that is responsible for the fluid temperature gradient near the surface. This dependence is contained in the convection heat transfer coefficient hc defined by @T q_ ¼ k hc ðT Ts Þ @n s 00

ð1:11Þ

16

INTRODUCTION TO FIRE

Convection is primarily given in terms of correlations of the form Nu

hc l ¼ C Ren Pr m k

or Nu

hc l ¼ C Gr n Pr k

ð1:12Þ

where the Grashof number, Gr, replaces the Re in the case of natural convection, which is common to most fire conditions. The Grashof number is defined as Gr

gðT T s Þl3

2

ð1:13Þ

where is the coefficient of volume expansion, 1=T for a gas, and Pr is the Prandtl number, cp =k. These correlations depend on the nature of the flow and the surface geometry. Mass transfer by laminar (molecular) diffusion is directly analogous to conduction with the analog of Fourier’s law as Fick’s law describing the mass flux (mass flow rate per unit area) of species i due to diffusion: i Vi ¼ m_ 00i ¼ D

@Yi @n

ð1:14Þ

where Vi is the diffusion velocity of the ith species, is the mixture density and Di is the diffusion coefficient and Yi is the mass fraction concentration of the ith species. The corresponding analog for mass flux in convection is given in terms of the mass transfer coefficient, hm , as m_ 00i ¼ hm ðYi Yi;s Þ

ð1:15Þ

The convective mass transfer coefficient hm can be obtained from correlations similar to those of heat transfer, i.e. Equation (1.12). The Nusselt number has the counterpart Sherwood number, Sh hm l=Di , and the counterpart of the Prandtl number is the Schmidt number, Sc ¼ =D. Since Pr Sc 0:7 for combustion gases, the Lewis number, Le ¼ Pr=Sc ¼ k=Dcp is approximately 1, and it can be shown that hm ¼ hc =cp . This is a convenient way to compute the mass transfer coefficient from heat transfer results. It comes from the Reynolds analogy, which shows the equivalence of heat transfer with its corresponding mass transfer configuration for Le ¼ 1. Fire involves both simultaneous heat and mass transfer, and therefore these relationships are important to have a complete understanding of the subject.

1.5.4

Supportive references

Standard texts on the prerequisites abound in the literature and any might serve as refreshers in these subjects. Other texts might serve to support the body of material in this text on fire. They include the following combustion and fire books: Cox, G. (ed.), Combustion Fundamentals of Fire, Academic Press, London, 1995. Drysdale, D., Introduction to Fire Dynamics, 1st edn., John Wiley & Sons, Ltd, Chichester, 1985. Glassman, I., Combustion, Academic Press, New York, 1977.

PROBLEMS

17

Kanury, A. M., Introduction to Combustion Phenomena, Gordon and Breach, New York, 1975. Karlsson B. and Quintiere J. G., Enclosure Fire Dynamics, CRC Press, Boca Raton, Florida, 2000. Turns, S. R., An Introduction to Combustion, 2nd edn, McGraw-Hill, Boston, 2000. Williams, F. A., Combustion Theory, 2nd edn, Addison-Wesley Publishing Co., New York, 1985.

References 1. Bland, R. E. (Chair), America Burning, The National Commission on Fire Prevention and Control, US Government Printing Office, Washington DC, 1973. 2. Brushlinski, N., Sokolov, S. and Wagner, P., World Fire Statistics at the End of the 20th Century, Report 6, Center of Fire Statistics of CTIF, International Technical Committee for the Prevention and Extinction of Fire, Moscow and Berlin, 2000. 3. Gomberg, A., Buchbinder, B. and Offensend, F. J., Evaluating Alternative Strategies for Reducing Residential Fire Loss – The Fire Loss Model, National Bureau of Standards, NBSIR 82-2551, August 1982. 4. The World’s Worst Disasters of the Twentieth Century, Octopus Books, London, 1999. 5. Sullivan, M. J., Property loss rises in large-loss fires, NFPA J., November/December 1994, 84–101. 6. Semenov, N., Chemical Kinetics and Chain Reactions, Moscow, 1934. 7. Frank-Kamenetskii, D. A., Diffusion and Heat Transfer in Chemical Kinetics, Plenum Press, New York, 1969. 8. Zel’dovich, Y. B., Barenblatt, G. I., Librovich, V. B. and Makhviladze, G. M., The Mathematical Theory of Combustion and Explosion, Nauka, Moscow, 1980. 9. Spalding, D. B., The combustion of liquid fuels, Proc. Comb. Inst., 1953, 4, pp. 847–64. 10. Emmons, H. W., The film combustion of liquid fuel, Z. Angew Math. Mech., 1956, 36, 60–71. 11. Cochrane, R. C., Measures for Progress – A History of the National Bureau of Standards, US Department of Commerce, 1966, p. 131. 12. Rasbash, D. J. and Alvarez, N. J. (eds), Advances in fire physics – a symposium to honor Philip H. Thomas, Fire Safety J., 1981, 3, 91–300.

Problems 1.1

You may have seen the History Channel documentary or other information pertaining to the making of the World Trade Towers. Write a short essay, no more than two pages, on how you would have protected the Towers from collapse. You are to use practical fire safety technology, supported by scientific reasoning and information. You need to specify clearly what you would have done to prevent the Towers from collapse, based on fire protection principles, and you must justify your position. You can use other sources, but do not prepare a literature report; I am looking for your thinking.

1.2

Why do fire disasters seem to repeat despite regulations developed to prevent them from occurring again? For example, contrast the recent Rhode Island nightclub fire with the infamous Coconut Grove and Beverly Hill Supper Club fires. Can you identify other fires in places of entertainment that were equally deadly?

1.3

Look up the significant people that contributed to the development of fire science. Pick one, and write a one-page description of the contribution and its significance.

18

INTRODUCTION TO FIRE

1.4

Examine the statistics of fire. Is there any logical reason why the US death rate is among the highest and why Russia’s increased so markedly after the fall of the Soviet Union?

1.5

The Oakland Hills fire in 1991 was a significant loss. Is the repetition of such fires due to earthquakes or the interface between the urban and wildlife domains? What is being done to mitigate or prevent these fires?

1.6

How does science enter into any of the codes and standards that you are familiar with?

2 Thermochemistry

2.1 Introduction Thermochemistry is the study of chemical reactions within the context of thermodynamics. Thermodynamics is the science of equilibrium states of matter which follow the conservation of mass and energy. Applications of the second law of thermodynamics can establish criteria for equilibrium and determine the possible changes of state. In general, changes of state take place over time, and states can vary over matter as spatial gradients. These gradients of states do not constitute equilibrium in the thermodynamic sense, yet it is justifiable to regard relationships between the thermodynamic properties to still apply at each point in a gradient field. This concept is regarded as local equilibrium. Thus the properties of temperature and pressure apply as well as their state relationships governed by thermodynamics. However, with regard to chemical properties, in application to fire and combustion, it is not usually practical to impose chemical equilibrium principles. Chemical equilibrium would impose a fixed distribution of mass among the species involved in the reaction. Because of the gradients in a real chemically reacting system, this equilibrium state is not reached. Hence, in practical applications, we will hypothesize the mass distribution of the reacting species or use measured data to prescribe it. The resultant states may not be consistent with chemical equilibrium, but all other aspects of thermodynamics will still apply. Therefore, we shall not include any further discussion of chemical equilibrium here, and the interested student is referred to any of the many excellent texts in thermodynamics that address this subject. Where applications to industrial combustion systems involve a relatively limited set of fuels, fire seeks anything that can burn. With the exception of industrial incineration, the fuels for fire are nearly boundless. Let us first consider fire as combustion in the gas phase, excluding surface oxidation in the following. For liquids, we must first require evaporation to the gas phase and for solids we must have a similar phase transition. In the former, pure evaporation is the change of phase of the substance without changing its composition. Evaporation follows local thermodynamics equilibrium between the gas Fundamentals of Fire Phenomena James G. Quintiere # 2006 John Wiley & Sons, Ltd ISBN: 0-470-09113-4

20

THERMOCHEMISTRY

and liquid phases. Heat is required to change the phase. Likewise, heat is required in the latter case for a solid, but now thermal decomposition is required. Except for a rare instance, of a subliming fuel, the fuel is not likely to retain much of its original composition. This thermal decomposition process is called pyrolysis, and the decomposed species can vary widely in kind and in number. Fuels encountered in fire can be natural or manmade, the latter producing the most complications. For example, plastics constitute a multitude of synthetic polymers. A polymer is a large molecule composed of a bonded chain of a repeated chemical structure (monomer). For example, the polymers polyethylene and polymethyl-methylmethracrylate (PMMA) are composed of the monomers C2 H4 and C5 H8 O2 respectively. The C2 H4 , standing alone, is ethylene, a gas at normal room temperatures. The properties of these two substances are markedly different, as contrasted by their phase change temperatures:

Melt temperatureð CÞ Vaporization temperatureð CÞ

Ethylene 169:1 103:7

Polyethylene 130 400

For ethylene, the phase transition temperatures are precise, while for polyethylene these are very approximate. Indeed, we may have pyrolysis occurring for the polymer, and the phase changes will not then be so well defined. Moreover, commercial polymers will contain additives for various purposes which will affect the phase change process. For some polymers, a melting transition may not occur, and instead the transition is more complex. For example, wood and paper pyrolyze to a char, tars and gases. Char is generally composed of a porous matrix of carbon, but hydrogen and other elements can be attached. Tars are high molecular weight compounds that retain a liquid structure under normal temperatures. The remaining products of the pyrolysis of wood are gases consisting of a mixture of hydrocarbons, some of which might condense to form an aerosol indicative of the ‘white smoke’ seen when wood is heated. The principal ingredient of wood is cellulose, a natural polymer whose monomer is C6 H10 O5 . Wood and other charring materials present a very complex path to a gaseous fuel. Also wood and other materials can absorb water from the atmosphere. This moisture can then change both the physical and chemical properties of the solid fuel. Thus, the pyrolysis process is not unique for materials, and their description for application to fire must be generally empirical or very specific.

2.2 Chemical Reactions Combustion reactions in fire involve oxygen for the most part represented as FuelðFÞ þ oxygenðO2 Þ ! productsðPÞ The oxygen will mainly be derived from air. The fuel will usually consist of mostly carbon (C), hydrogen (H) and oxygen (O) atoms in a general molecular structure, F : Cx Hy Oz . Fuels could also contain nitrogen (N), e.g. polyurethane, or chlorine (Cl), e.g. polyvinylchloride ðC2 H3 ClÞn .

CHEMICAL REACTIONS

21

We use F as a representative molecular structure of the fuel in terms of its atoms and P, a similar description for the product. Of course, we can have more than one product, but symbolically we only need to represent one here. The chemical reaction can then be described by the chemical equation as F F þ 0 O 2 þ D D ! P P þ D D

ð2:1Þ

where the ’s are the stoichiometric coefficients. The stoichiometric coefficients are determined to conserve atoms between the left side of the equation, composing the reactants, and the right side, the products. In general, P P represents a sum over P;i Pi for each product species, i. Fuel or oxygen could be left over and included in the products, as well as an inert diluent (D) such as nitrogen in air, which is carried without change from the reactants to the products. A complete chemical reaction is one in which the products are in their most stable state. Such a reaction is rare, but, in general, departure from it is small for combustion systems. For example, for the fuel Cx Hy Oz , the complete products of combustion are CO2 and H2 O. Departures from completeness can lead to additional incomplete products such as carbon monoxide (CO), hydrogen ðH2 Þ and soot (mostly C). For fuels containing N or Cl, incomplete products of combustion are more likely in fire, yielding hydrogen cyanide (HCN) and hydrogen chloride (HCl) gases instead of the corresponding stable products: N2 and Cl2 . In fire, we must rely on measurements to determine the incompleteness of the reaction. Where we might ignore these incomplete effects for thermal considerations, we cannot ignore their effects on toxicity and corrosivity of the combustion products. A complete chemical reaction in which no fuel and no oxygen is left is called a stoichiometric reaction. This is used as a reference, and its corresponding stoichiometric oxygen to fuel mass ratio, r, can be determined from the chemical equation. A useful parameter to describe the state of the reactant mixture is the equivalence ratio, , defined as mass of available fuel ¼ r ð2:2Þ mass of available oxygen If < 1, we have burned all of the available fuel and have leftover oxygen. This state is commonly called ‘fuel-lean’. On the other hand, if > 1, there is unburned gaseous fuel and all of the oxygen is consumed. This state is commonly called ‘fuel-rich’. When we have a fire beginning within a room, is less than 1, starting out as zero. As the fire grows, the fuel release can exceed the available oxygen supply. Room fires are termed ventilation-limited when 1. The chemical equation described in Equation (2.1) can be thought of in terms of molecules for each species or in terms of mole ðnÞ. One mole is defined as the mass ðmÞ of the species equal to its molecular weight ðMÞ. Strictly, a mole (or mol) is a gram-mole (gmole) in which m is given in grams. We could similarly define a pound-mole (lb-mole), which would contain a different mass than a gmole. The molecular weight is defined as the sum of the atomic weights of each element in the compound. The atomic weights have been decided as a form of relative masses of the atoms with carbon (C) assigned the value of 12, corresponding to six protons and six neutrons.* Thus, a mole is a relative * The average atomic weight for carbon is 12.011 because about 1 % of carbon found in nature is the isotope, carbon 13, having an extra neutron in the nucleus.

22

THERMOCHEMISTRY

mass of the molecules of that species, and a mole and molecular mass are proportional. The constant of proportionality is Avogadro’s number, N0 , which is 6:022 1023 molecules=mole. For example, the mass of a methane molecule is determined from CH4 using the atomic weights C ¼ 12:011 and H ¼ 1:007 94: ½ð1Þð12:011Þ þ 4ð1:007 94Þ

g 1 mole g

¼ 2:66 1023 mole 6:022 1023 molecule molecule

The chemical equation then represents a conservation of atoms, which ensures conservation of mass and an alternative view of the species as molecules or moles. The stoichiometric coefficients correspond to the number of molecules or moles of each species. Example 2.1 Ten grams of methane ðCH4 Þ are reacted with 120 g of O2 . Assume that the reaction is complete. Determine the mass of each species in the product mixture. Solution

The chemical equation is for the stoichiometric reaction: CH4 þ 0 O2 ! P1 CO2 þ P2 H2 O

where F was taken as 1. Thus, Conserving C atoms : Conserving H atoms : Conserving O atoms :

P1 ¼ 1 4 ¼ 2P2 or P2 ¼ 2 20 ¼ 2 þ 2 or 0 ¼ 2

Hence, CH4 þ 2 O2 ! CO2 þ 2 H2 O as the stoichiometric reaction. The reactants contain 10 g of CH4 and 120 g of O2 . The stoichiometric ratio is r¼

mO2 mCH4

for the stoichiometric reaction

Since the stoichiometric reaction has 1 mole (or molecule) of CH4 and 2 moles of O2 , and the corresponding molecular weights are MCH4 ¼ 12 þ 4ð1Þ ¼ 16 g=mole as C has an approximate weight of 12 and H is 1, and MO2 ¼ 2ð16Þ ¼ 32 g=mole we compute r¼

ð2 moles O2 Þð32 g=mole O2 Þ ¼ 4 g O2 =g CH4 ð1 mole CH4 Þð16 g=mole CH4 Þ

GAS MIXTURES

23

The equivalence ratio is then ¼

10 g CH4 1 ð4Þ ¼ < 1 3 120 g O2

Therefore, we will have O2 left in the products. In the products, we have mO2 ¼ 120 g mO2 ;reacted

mCO2 mH2 O

4g O2 ¼ 120 g ð10 g CH4 Þ ¼ 80 g O2 1 g CH4 1 mole CO2 44g=mole CO2 ¼ ð10 g CH4 Þ ¼ 27:5 g CO2 1 mole CH4 16 g=mole CH4 2 moles H2 O 18 g=mole H2 O ¼ ð10 g CH4 Þ ¼ 22:5 g H2 O 1 mole CH4 16 g=mole CH4

Note that mass is conserved as 130 g was in the reactant as well as the product mixture.

2.3 Gas Mixtures Since we are mainly interested in combustion in the gas phase, we must be able to describe reacting gas mixtures. For a thermodynamic mixture, each species fills the complete volume ðVÞ of the mixture. For a mixture of N species, the mixture density ðÞ is related to the individual species densities ði Þ by ¼

N X

i

ð2:3Þ

i¼1

Alternative ways to describe the distribution of species in the mixture are by mass fraction, Yi ¼

i mi =V mi ¼ ¼ m=V m

ð2:4Þ

ni n

ð2:5Þ

and by mole fraction, Xi ¼

where n is the sum of the moles for each species in the mixture. By the definition of molecular weight, the molecular weight of a mixture is M¼

N X i¼1

Xi M i

ð2:6Þ

24

THERMOCHEMISTRY

It follows from Equations (2.4) and (2.6) that the mass and mole fractions are related as Yi ¼

n i M i Xi M i ¼ nM M

ð2:7Þ

Example 2.2 Air has the approximate composition by mole fraction of 0:21 for oxygen and 0:79 for nitrogen. What is the molecular weight of air and what is its corresponding composition in mass fractions? Solution

By Equation (2.6), moles O2 Ma ¼ 0:21 ð32 g=mole O2 Þ mole air moles N 2 þ 0:79 ð28 g=mole N2 Þ mole air Ma ¼ 28:84 g=mole air

By Equation (2.7), YO2 ¼

ð0:21Þð32Þ ¼ 0:233 ð28:84Þ

and YN2 ¼ 1 YO2 ¼ 0:767 The mixture we have just described, even with a chemical reaction, must obey thermodynamic relationships (except perhaps requirements of chemical equilibrium). Thermodynamic properties such as temperature ðTÞ, pressure ðpÞ and density apply at each point in the system, even with gradients. Also, even at a point in the mixture we do not lose the macroscopic identity of a continuum so that the point retains the character of the mixture. However, at a point or infinitesimal mixture volume, each species has the same temperature according to thermal equilibrium. It is convenient to represent a mixture of gases as that of perfect gases. A perfect gas is defined as following the relationship p¼

nRT V

where R is the universal gas constant, 8:314 J=mole K It can be shown that specific heats for a perfect gas only depend on temperature.

ð2:8Þ

CONSERVATION LAWS FOR SYSTEMS

25

A consequence of mechanical equilibrium in a perfect gas mixture is that the pressures developed by each species sum to give the mixture pressure. This is known as Dalton’s law, with the species pressure called the partial pressure, pi : p¼

N X

pi

ð2:9Þ

i¼1

Since each species is a perfect gas for the same volume ðVÞ, it follows that pi ni RT=V ni ¼ ¼ Xi ¼ nRT=V p n

ð2:10Þ

This is an important relationship between the mole fraction and the ratio of partial and mixture pressures. Sometimes mole fractions are called volume fractions. The volume that species i would occupy if allowed to equilibrate to the mixture pressure is Vi ¼

ni RT p

From Equation (2.8) we obtain Vi n i ¼ ¼ Xi V n Hence, the name volume fraction applies.

2.4 Conservation Laws for Systems A system is considered to be a collection of matter fixed in mass. It can exchange heat and work with its surroundings, but not mass. The conservation laws are expressed for systems and must be adjusted when the flow of matter is permitted. We will return to the flow case in Chapter 3, but for now we only consider systems. The conservation of mass is trivial to express for a system since the mass ðmÞ is always fixed. For a change in the system from state 1 to state 2, this is expressed as m1 ¼ m2

ð2:11aÞ

dm ¼0 dt

ð2:11bÞ

or considering variations over time,

Alternatively, the rate or time-integrated form applies to the system.

26

THERMOCHEMISTRY

The conservation of momentum or Newton’s second law of motion is expressed for a system as dðmvÞ dt

F¼

ð2:12Þ

where F is the vector sum of forces acting on the system and v is the velocity of the system. If v varies over the domain of the system, we must integrate v over the system volume before we take the derivative with time. Implicit in the second law is that v must be measured with respect to a ‘fixed’ reference frame, such as the Earth. This is called an inertial frame of reference. Although we will not use this law often in this book, it will be important in the study of fire plumes in Chapter 10. Since our system contains reacting species we need a conservative law for each species. Consider the mass of species iðmi Þ as it reacts from its reactant state (1) to its product state (2). Then since the mass of i can change due to the chemical reaction with mi;r being produced by the reaction mi;2 ¼ mi;1 þ mi;r

ð2:13Þ

dmi ¼ m_ i;r dt

ð2:14Þ

or as a rate

The right-hand side of Equation (2.14) is the mass rate of production of species i due to the chemical reaction. This term represents a generation of mass. However, for the fuel or oxygen it can be a sink, having a negative sign. From Example 2.1, mO2 ;2 is 80 g, mO2 ;1 is 120 g and mO2 ;r is 40 g. The conservation of energy or the first law of thermodynamics is expressed as the change in total energy of the system, which is equal to the net heat added to the system from the surroundings ðQÞ minus the net work done by the system on the surroundings ðWÞ. Energy ðEÞ is composed of kinetic energy due to macroscopic motion and internal energy ðUÞ due to microscopic effects. These quantities, given as capital letters, are extensive properties and depend on the mass of the system. Since each species in a mixture contributes to U we have for a system of volume ðVÞ,

U¼V

N X

i ui

ð2:15Þ

i¼1

where ui is the intensive internal energy given as Ui =mi . Since the mass of the system is fixed from Equations (2.4) and (2.5), it can be shown that

u¼

N X i¼1

Yi ui

ð2:16Þ

CONSERVATION LAWS FOR SYSTEMS

27

or ~ u¼

N X

Xi ~ ui

ð2:17Þ

i¼1

where the designation ð~Þ means per mole. Furthermore, since u and u~ are intensive properties valid at a point, the relationships given in Equations (2.16) and (2.17) always hold at each point in a continuum. In all our applications, kinetic energy effects will be negligible, so an adequate expression for the first law is U2 U1 ¼ Q W

2.4.1

ð2:18Þ

Constant pressure reaction

For systems with a chemical reaction, an important consideration is reactions that occur at constant pressure. This could represent a reaction in the atmosphere, such as a fire, in which the system is allowed to expand or contract according to the pressure of the surrounding atmosphere. Figure 2.1 illustrates this process for a system contained in a cylinder and also bounded by a frictionless and massless piston allowed to move so that the pressure is always constant on each side. We consider the work term to be composed of that pertaining to forces associated with turning ‘paddlewheel’ shafts and shear stress ðWs Þ, and work associated with normal pressure forces ðWp Þ. For most applications in fire, Ws will not apply and therefore we will ignore it here. For the piston of face area, A, the work due to pressure ðpÞ on the surroundings, moving at a distance ðx2 x1 Þ, is Wp ¼ pAðx2 x1 Þ

Figure 2.1

Constant pressure process

28

THERMOCHEMISTRY

Since the volume of the system can be expressed by Ax, Wp ¼ pðV2 V1 Þ or W p ¼ p2 V2 p 1 V1 since p1 ¼ p2 ¼ p. Then the first law (Equation (2.18)) becomes U2 U1 ¼ Q ðp2 V2 p1 V1 Þ The property enthalpy ðHÞ, defined as H ¼ U þ pV

ð2:19Þ

is now introduced. Thus, for a constant pressure reacting system, without shaft or shear work, we have H2 H1 ¼ Q

2.4.2

(2.20)

Heat of combustion

In a constant pressure combustion system in which state 1 represents the reactants and state 2 the products, we expect heat to be given to the surroundings. Therefore, Q is negative and so is the change in enthalpy according to Equation (2.20). A useful property of the reaction is the heat of combustion, which is related to this enthalpy change. We define the heat of combustion as the positive value of this enthalpy change per unit mass or mole of fuel reacted at 1 atm and in which the temperature of the system before and after the reaction is 25 C. It is given as hc ¼

ðH2 H1 Þ mF;r

ð2:21aÞ

~ hc ¼

ðH2 H1 Þ nF;r

ð2:21bÞ

and

with H2 and H1 evaluated at 1 atm, 25 C. Often the heat of combustion is more restrictive in its definition, applying only to the stoichiometric (complete) reaction of the fuel. Sometimes the heat of combustion is given as a negative quantity which applies to Equation (2.21) without the minus sign before the parentheses. Since combustion gives rise to heat transfer, it is more rational to define it as positive. Also it should become apparent that the heat of combustion represents the contribution of energy from the sole process of rearranging the atoms into new molecules according to the chemical equation.

CONSERVATION LAWS FOR SYSTEMS

Figure 2.2

29

Idealized oxygen bomb calorimeter

In other words, there is no energy contribution due to a temperature change since it remains constant at 25 C. The heat of combustion of solids or liquids is usually measured in a device known as an oxygen bomb calorimeter. Such a device operates at a constant volume between states 1 and 2, and its heat loss is measured by means of the temperature rise to a surrounding water-bath. This is schematically shown in Figure 2.2. The combustion volume is charged with oxygen and a special fuel is added to ensure complete combustion of the fuel to be measured. Since the process is at constant volume ðVÞ, we have U2 U1 ¼ Q from Equation (2.18). By the definition of enthalpy, we can write H2 H1 ¼ Q þ ðp2 p1 ÞV By analysis and by measurement, we can determine H2 H1 and express it in terms of values at 25 C and 1 atm. Since Q is large for combustion reactions, these pressure corrections usually have a small effect on the accuracy of the heat of combustion determined in this way.

2.4.3

Adiabatic flame temperature

The adiabatic flame temperature is defined as the maximum possible temperature achieved by the reaction in a constant pressure process. It is usually based on the reactants initially at the standard state of 25 C and 1 atm. From Equation (2.20), the adiabatic temperature ðTad Þ is determined from the state of the system at state 2 such that H2 ðTad Þ ¼ H1 ð25 C and 1 atmÞ

ð2:22Þ

We shall now see how we can express H2 in terms of the product mixture. It might be already intuitive, but we should realize that Tad depends on the state of the reactants. In addition to the fuel, it depends on whether oxygen is left over, and if a diluent, such as nitrogen, must also be heated to state 2. Reactants that remain in the product state reduce

30

THERMOCHEMISTRY

Tad since the energy associated with the combustion has been used to heat these species along with the new ones produced.

2.5 Heat of Formation The molar heat of formation is the heat required to produce one mole of a substance from its elemental components at a fixed temperature and pressure. The fixed temperature and pressure is called the standard state and is taken at 25 C and 1 atm. The heat of formation is designated by the symbol hf;i for the ith species and is given in energy units per unit mass or mole of species i. Usually, substances composed of a single element have hf;i values at zero, but this is not always true since it depends on their state. For example, carbon as graphite is 0, but carbon as a diamond is 1.88 kJ/mole. Diamond has a different crystalline structure than the graphite, and thus it takes 1.88 kJ/mole to produce 1 mole of C (diamond) from 1 mole of C (graphite). This is a physical, not chemical, change of the molecular structure. Again, we are complying with the first law of thermodynamics in this process, but we do not know if or how the process can proceed. An alternative, but equivalent, way to define the heat of formation of a species is to define it as the reference enthalpy at the standard state ð25 C; 1 atmÞ. Thus, hi ð25 C; 1 atmÞ hf;i

ð2:23Þ

Strictly, this definition only applies to substances in equilibrium at the standard state. For a perfect gas, and approximately for solids and liquids at small changes from 1 atm pressure, the enthalpy is only a function of temperature. It can be written in terms of specific heat at constant pressure, cp , as hi ðTÞ ¼ hf;i þ

ðT cp;i dT

(2.24)

25 C

Strictly, this definition only applies to substances in equilibrium at the standard state. For the high temperatures achieved in combustion, cp should always be regarded as a function of temperature. Some values are shown in Table 2.1. Values for the molar heats of formation for several substances are listed in Table 2.2. More extensive listings have been compiled and are available in reference books. The student should recognize that the values have been determined from measurements and from the application of Equation (2.20). Some examples should show the utility and interpretation of the heat of formation. Table 2.1

Specific heat at constant pressure, ~cp;i ðJ=K moleÞ (abstracted from Reference [1])

Temperature ðKÞ H2 ðgÞ 298 600 1000 1600 2500

28.8 29.3 30.2 32.7 35.8

CðsÞ

CH4 ðgÞ

8.5 16.8 21.6 24.2 25.9

35.8 52.2 71.8 88.5 98.8

(g) gaseous state, (s) graphite, solid state.

C2 H4 ðgÞ

O2 ðgÞ

N2 ðgÞ

H2 OðgÞ

CO2 ðgÞ

COðgÞ

42.9 70.7 110.0 112.1 123.1

29.4 32.1 34.9 36.8 38.9

29.1 30.1 32.7 35.1 36.6

33.6 36.3 41.3 48.1 53.9

37.1 47.3 54.3 58.9 61.5

29.1 30.4 33.2 35.5 36.8

APPLICATION OF MASS AND ENERGY CONSERVATION

31

Table 2.2 Heat of formation ~hf in kJ=mole (at 25 C and 1 atm)a (abstracted from Reference [2]) Substance

Formula

Oxygen Nitrogen Graphite Diamond Carbon dioxide Carbon monoxide Hydrogen Water Water Chlorine Hydrogen chloride Hydrogen cyanide Methane Propane n-Butane n-Heptane Benzene Formaldehyde Methanol Methanol Ethanol Ethylene

O2 N2 C C CO2 CO H2 H2 O H2 O Cl2 HCl HCN CH4 C3 H8 C4 H10 C7 H16 C6 H6 CH2 O CH4 O CH4 O C2 H6 O C2 H4

State

~ (kJ/mole) h f

g g s s g g g g l g g g g g g g g g g l l g

0 0 0 1.88 393.5 110.5 0 241.8 285.9 0 92.3 þ135.1 74.9 103.8 124.7 187.8 þ82.9 115.9 201.2 238.6 277.7 52.5

a

Values for gaseous substances not in equilibrum at the standard state have been determined from the liquid and the heat of vaporization.

2.6 Application of Mass and Energy Conservation in Chemical Reactions Example 2.3 Consider the formation of benzene and carbon dioxide from their elemental substances. Find the heat required per unit mole at the standard state: 25 C, 1 atm. Solution

For benzene, formed from graphite and hydrogen gas, 6 C þ 3 H2 ! C6 H6

is the chemical equation. By the application of Equation (2.20), we express the total enthalpies in molar form as e 1 ¼ ð6 moles CÞðe H hC ð25 CÞÞ þ ð3 moles H2 Þðe hH2 ð25 CÞÞ and e 2 ¼ ð1 mole C6 H6 Þðe H hC6 H6 ð25 CÞÞ

32

THERMOCHEMISTRY

Since the enthalpies are the heats of formation by Equation (2.23), we find from Table 2.2 that Q ¼ e hf for benzene or Q ¼ þ82:9 kJ=mole of benzene and heat is required. In a similar fashion, obtaining carbon dioxide from graphite (carbon) and oxygen C þ O2 ! CO2 gives 393.5 kJ/mole for CO2 for Q, or heat is lost. This reaction can also be viewed as the combustion of carbon. Then by Equation (2.21b), this value taken as positive is exactly the heat of combustion of carbon; i.e. e hc ¼ 393:5 kJ=mole carbon. Example 2.4 Suppose we have the stoichiometric reaction for carbon leading to CO2 , but in the final state after the reaction the temperature increased from 25 to 500 C. Find the heat lost. Solution

Since state 1 is at 25 C for C and O2 , we obtain as before e1 ¼ 0 H

However, at state 2, we write, from Equation (2.24), ð 500 C e 2 ¼ n2 e H cp dT h2 ¼ ð1 mole CO2 Þ hf þ 25 C

From Tables 2.1 and 2.2, we take a mean value of e cp ¼ 47 J=mole K and obtain e 2 ¼ ð1 moleÞ 393:5 kJ=mole þ 47 103 kJ=mole Kð475 KÞ H or e 2 ¼ 371:2 kJ=mole H Therefore, the heat lost is 371.2 kJ/mole. Example 2.5 One mole of CO burns in air to form CO2 in a stoichiometric reaction. Determine the heat lost if the initial and final states are at 25 C and 1 atm. Solution

The stoichiometric reaction is CO þ air ! CO2 þ N2

APPLICATION OF MASS AND ENERGY CONSERVATION

33

where all of the CO and all of the O2 in the air are consumed. It is convenient to represent air as composed of 1 mole O2 þ 3:76 moles of N2 , which gives 4.76 moles of the mixture, air. These proportions are the same as XO2 ¼ 0:21 and XN2 ¼ 0:79. The chemical equation is then CO þ

1 3:76 ðO2 þ 3:76 N2 Þ ! CO2 þ N2 2 2

We designate state 1 as reactants and state 2 as the product mixture. By Equation (2.20), Q ¼ H2 H1 Using molar enthalpies, we develop the total enthalpy of each mixture as H¼

N X

ni e hi

i¼1

where e hi is found from Equation (2.24) and using Tables 2.1 and 2.2, 1 H1 ¼ ð1 mole COÞð110:5 kJ=moleÞ þ ð0Þ ¼ 110:5 kJ 2 3:76 H2 ¼ ð1 mole CO2 Þð393:5 kJ=moleÞ þ ð0Þ ¼ 393:5 kJ 2 Q ¼ 393:5 ð110:5Þ ¼ 283:0 kJ From our definition of the heat of combustion, the result in Example 2.5 is hc ¼ unequivocally the heat of combustion of CO, b hc ¼ 283 kJ=mole CO, or e 283=ð12 þ 16Þ ¼ 10:1 kJ=g CO. From Equations (2.21) and (2.24), we can generalize this result as e hc ¼

N X

! i e hf;i

i¼1

N X i¼1

Reactants

!

i e hf;i

ð2:25Þ Products

where the i ’s are the molar stoichiometric coefficients with F is taken as one. Usually, tabulated values of hc for fuels are given for a stoichiometric (complete) reaction. A distinction is sometimes given between H2 O being a liquid in the product state (‘gross’ value) or a gas (‘net’ value). The gross heat of combustion is higher than the net because some of the chemical energy has been used to vaporize the water. Indeed, the difference between the heats of formation of water in Table 2.2 for a liquid (285.9 kJ/mole) and a gas (241/8 kJ/mole) is the heat of vaporization at 1 atm and 25 C: hv ¼

ð285:9 241:8Þ kJ=mole ¼ 2:45 kJ=g H2 O 18 g=mole H2 O

34

THERMOCHEMISTRY

Example 2.6 Consider the same problem as in Example 2.5, but now with the products at a temperature of 625 C. Solution To avoid accounting for cp variations with temperature, we assume constant average values. We select approximate average values from Table 2.1 as cp ;N2 ¼ 30 J=mole N2 K cp ;CO2 ¼ 47 J=mole CO2 K Then for H2 we write H2 ¼

2 X i hf;i þ cpi ðT2 25 CÞ i¼1

where we sum over the products CO2 and N2 : H2 ¼ ð1 mole CO2 Þ 393:5 kJ=mole CO2 þ 0:047 kJ=mole Kð600 KÞ

3:76 moles N2 0 þ 0:030 kJ=mole Kð600 KÞ þ 2 3:76 ¼ ð393:5Þ þ ð1Þð0:047Þ þ ð0:030Þ ð600Þ 2 ¼ ð393:5Þ þ ð0:10325Þð600Þ ¼ 331:55 kJ Q ¼ H2 H1 ¼ ð331:55Þ ð110:5Þ ¼ 221:05 kJ For Example 2.6 we see that the heat lost due to the chemical reaction is composed of two parts: one comes from the heats of formation and the other from energy stored in the products mixture to raise its temperature. It can be inferred from these examples that we can write

ðQÞ ¼ F e hc þ

N;Reactants X i¼1

N;Products X i¼1

ni e cpi ðT1 25 CÞ ð2:26Þ

nie cpi ðT2 25 CÞ

where F is the moles (or mass) of fuel reacted and ni are the moles (or mass) of each species in the mixture before (1, reactants) and after (2, products) the reaction. If we

COMBUSTION PRODUCTS IN FIRE

35

further consider specific heat of the mixture, since from Equations (2.17), (2.19), (2.24) and (2.25) e cp ¼

@e h @T

! ¼ p

X @ X Xi h i ¼ cpi Xie @T

X 1 ni e ¼ cpi n

ð2:27Þ

then ðQÞ ¼ F e hc þ n1e cp1 ðT1 25 CÞ n2e cp2 ðT2 25 CÞ

ð2:28Þ

In other words, the heat lost in the chemical reaction is given by a chemical energy release plus a ‘sensible’ energy associated with the heat capacity and temperature change of the mixture states.

2.7 Combustion Products in Fire Up to now we have only considered prescribed reactions. Given a reaction, the tools of thermodynamics can give us the heat of combustion and other information. In a combustion reaction we could impose conditions of chemical equilibrium, or ideally complete combustion. While these approximations can be useful, for actual fire processes we must rely on experimental data for the reaction. The interaction of turbulence and temperature variations can lead to incomplete products of combustion. For fuels involving Cx Hy Oz we might expect that Cx Hy Oz þ O2 ! CO2 ; CO; H2 O; H2 ; soot; CH where soot is mostly carbon and by CH we mean the residual of hydrocarbons. Unless one can show a chemical similarity among the various burning conditions, the combustion products will be dependent on the experimental conditions. Tewarson [3] has studied the burning characteristics of a wide range of materials. He has carried out experiments mainly for horizontal samples of 10 cm 10 cm under controlled radiant heating and air supply conditions. Chemical measurements to determine the species in the exhaust gases have identified the principal products of combustion. Measurement of the mass loss of the sample during combustion will yield the mass of gas evolved from the burning sample. This gas may not entirely be fuel since it could contain evaporated water or inert fire retardants. It is convenient to express the products of combustion, including the chemical energy liberated, in terms of this mass loss. This is defined as a yield, i.e. yield of combustion product i; yi

mass of species i mass loss of sample

36

THERMOCHEMISTRY

Values of yields for various fuels are listed in Table 2.3. We see that even burning a pure gaseous fuel as butane in air, the combustion is not complete with some carbon monoxide, soot and other hydrocarbons found in the products of combustion. Due to the incompleteness of combustion the ‘actual’ heat of combustion (42.6 kJ/g) is less than the ‘ideal’ value (45.4 kJ/g) for complete combustion to carbon dioxide and water. Note that although the heats of combustion can range from about 10 to 50 kJ/g, the values expressed in terms of oxygen consumed in the reaction ðhO2 Þ are fairly constant at 13:0 0:3 kJ/g O2 . For charring materials such as wood, the difference between the actual and ideal heats of combustion are due to distinctions in the combustion of the volatiles and subsequent oxidation of the char, as well as due to incomplete combustion. For example, Wood þ heat ! volatiles þ char Volatiles þ air ! incomplete products; hc;vol Char þ air ! incomplete products; hc;char hc ;wood ¼ hc;vol þ hc ;char where here the heats of combustion should all be considered expressed in terms of the mass of the original wood. For fuels that completely gasify to the same chemical state as in a phase change for a pure substance, the yield is equivalent to the stoichiometric coefficient. However, for a fuel such as wood, which only partially gasifies and the gaseous products are no longer the original wood chemical, we cannot equate yield to a stoichiometric coefficient. The use of ‘yield’ is both practical and necessary in characterizing such complex fuels. Let us examine the measured benzene reaction given in Table 2.3. From the ideal complete stoichiometric reaction we have C6 H6 þ

15 O2 ! 6 CO2 þ 3 H2 O 2

The ideal heat of combustion is computed from Equation (2.25) using Table 2.2 as the net value (H2 O as a gas) ~ hc ¼ ½ð1Þðþ82:9Þ ½ð6Þð393:5Þ þ ð3Þð241:8Þ ¼ 3169:3 kJ=mole C6 H6 or hc ¼ ð3169:3Þ=ð78g=moleÞ ¼ 40:6 kJ=g This is higher than the value given in Table 2.3 (i.e., 40.2 kJ/g) since we used benzene from Table 2.2 as a gas and here it was burned as a liquid. The difference in energy went into vaporizing the liquid benzene and was approximately 0.4 kJ/g. (See Table 6.1)

Formula

CH4 C3 H8 C4 H10 CH4 O C2 H6 O C7 H16 C6 H6 CH1:7 O0:72 (C6 H11 NOÞn ðC5 H8 O2 Þn ðC2 H4 Þn ðC3 H6 Þn ðC8 H8 Þn

Substance

Methane (g) Propane (g) Butane (g) Methanol (l) Ethanol (l) n-Heptane (l) Benzene (g) Wood (red oak) Nylon PMMA Polyethyene, PE Polypropylene, PS Polystyrene, PS

Table 2.3

50.1 46.0 45.4 20.0 27.7 44.6 40.2 17.1 30.8 25.2 43.6 43.4 39.2

Ideal ðkJ=g fuel burnedÞ 49.6 43.7 42.6 19.1 25.6 41.2 27.6 12.4 27.1 24.2 30.8 38.6 27.0

Actual ðkJ=g mass lossÞ

Heat of combustion hc

12.5 12.9 12.6 13.4 13.2 12.7 13.0 13.2 11.9 13.1 12.8 12.7 12.7

Heat of combustion per mass of O2 hO2 ðkJ=g O2 consumedÞ 2.72 2.85 2.85 1.31 1.77 2.85 2.33 1.27 2.06 2.12 2.76 2.79 2.33

CO2 — 0.005 0.007 0.001 0.001 0.010 0.067 0.004 0.038 0.010 0.024 0.024 0.060

CO

— 0.024 0.029 — 0.008 0.037 0.181 0.015 0.075 0.022 0.060 0.059 0.164

Soot

— 0.003 0.003 — 0.001 0.004 0.018 0.001 0.016 0.001 0.007 0.006 0.014

Other hydrocarbon

Actual yields (g products/g mass lost)

Products of combustion in fire with sufficient air (abstracted from Tewarson [3])

38

THERMOCHEMISTRY

Let us continue further with this type of analysis, but now consider the actual reaction measured. We treat soot as pure carbon and represent the residual hydrocarbons as Cx Hy . Then the equation, without the proper coefficients, is C6 H6 þ O2 O2 ! CO2 ; CO; H2 O; C; Cx Hy Since benzene is a pure substance and we continue to treat it as a gas, we can regard the yields in this case as stoichiometric coefficients. From the measured yields in Table 2.3 and the molecular weights

Mi (g/mole)

C6H6

O2

CO2

CO

C

CxHy

78

32

44

28

12

12x + y

We calculate the stoichiometric efficients, i : CO2 : CO : C:

78 2:33gCO2 =gC6 H6 ¼ 44 78 ¼ ð0:067Þ 28 78 ¼ ð0:181Þ 12 Total

4:13 0:187 1:177 5:494 5:49

Since we must have six carbon atoms, we estimate x ¼ 0:51 and balance the H and O atoms: H: O:

6 ¼ 2H2 O þ y 2O2 ¼ ð2Þð4:13Þ þ ð0:187Þ þ H2 O ð1Þ

Selecting arbitrarily y ¼ 1, we obtain C6 H6 þ 5:47 O2 ! 4:13 CO2 þ 0:187 CO þ 1:18 C þ 2:5 H2 O þ C0:51 H Since we do not know the composition of the hydrocarbons, we cannot deal correctly with the equivalent compound C0:51 H. This is likely to be a mixture of many hydrocarbons including H2 . However, for purposes of including its effect on the actual heat of combustion, we will regard it as 14 C2 H4 , ethylene, as an approximation. Then, as before, we compute the heat of combustion: ~ hc ¼ ½ð1Þðþ82:9Þ ½ð4:13Þð393:5Þ þ ð0:187Þð110:5Þ þ ð1:18Þð0Þ þ ð2:5Þð241:8Þ þ ð0:25Þðþ52:5Þ ~ hc ¼ 2320:1 kJ=mole C6 H6

COMBUSTION PRODUCTS IN FIRE

39

or ~ hc ¼ 29:7 kJ=g C6 H6 This is very close to the value given in Table 2.3 determined by measurements (i.e. 27.6 kJ/g). One method of obtaining the chemical energy release in a fire is to measure the amount of oxygen used. Alternatively, some have considered basing this on the CO2 and CO produced. Both methods are based on the interesting fact that calculation of the heat of combustion based on the oxygen consumed, or alternatively the CO2 and CO produced, is nearly invariant over a wide range of materials. Huggett [4] demonstrated this for typical fire reactions, and Table 2.3 illustrates its typical variation. For the stoichiometric benzene reaction, it is listed in Table 2.3 as 13.0 kJ/g O2 . For the actual benzene reaction from our estimated result, we would obtain

hO2

78 g C6 H6 ¼ ð29:7 kJ=g C6 H6 Þ ð5:47Þð32Þ g O2

¼ 13:2 kJ=g O2

This result illustrates that even for realistic reactions, the hO2 appears to remain nearly constant. Table 2.3 data appear to remain invariant as long as there is sufficient air. By this we mean that with respect to the stoichiometric reaction there is more air than required. In terms of the equivalence ratio, we should have < 1. When is zero we have all oxygen and no fuel and when ¼ 1 we have the stoichiometric case with no oxygen or fuel in the products. However, when > 1 we see large departures to the data in Table 2.3. In general, CO soot and hydrocarbons increase; hc accordingly decrease. Tewarson [3] vividly shows in Figures 2.3 and 2.4 how the hc and the yield of CO changes for six

Figure 2.3

Actual heat of combustion in terms of the equivalence ratio where hc ;1 is the reference value given in Table 2.3 [3]

40

THERMOCHEMISTRY

Figure 2.4

Yield of CO in terms of the equivalence ratio where the normalize yield, yCO;1 , is given in Table 2.3 [3]

realistic fuels listed in Table 2.3. These results have importance in modeling ventilationlimited fires and determining their toxic hazard. These data are from the FMRC flammability apparatus [3] using the 10 cm 10 cm horizontal sample, and for enclosure fires [5]. The flaming results extend to ¼ 4 in Figures 2.3 and 2.4, at which point gas phase combustion appears to cease. However, combustion must continue since the heat of combustion remains nonzero. This is due to oxidation of the remaining solid fuel. If we consider wood, it would be the oxidation of the surface char composed primarily of carbon. From Example 2.3, we obtain the heat of combustion for carbon (going to CO2 ) as 32.8 kJ/g carbon. From Figure 2.4, we see a significant production of carbon monoxide at ¼ 4, and therefore it is understandable that Figure 2.3 yields a lower value: hc =hc ;1 ¼ 0:4 with hc ;1 from Table 2.3 of 12.4 kJ/g, or hc at ¼ 4 is 4.96 kJ/g. Here the actual heat of combustion of the char (under nonflaming conditions) at ¼ 4 is 4.96 kJ/g of mass lost while its ideal value is 32.8 kJ/g of carbon reacted. The ‘actual’ value is not based on stoichemistry of the fuel burned, but on the mass evolved to the gaseous state. The mass evolved may not be equal to the mass of fuel burned. Therefore, we have shown that thermodynamic principles and properties can be used to describe combustion reactions provided their stoichiometry is known. Since measurements in fire are based on the fuel mass lost, yields are used as empirical properties to describe the reaction and its products. When fire conditions become ventilation-limited ð 1Þ, the yield properties of a given fuel depend on . Although the generality of the results typified by Figures 2.3 and 2.4 have not been established, their general trends are accepted.

PROBLEMS

41

References 1. JANNAF Thermochemical Tables, 3rd edn., American Institute of Physics for the National Bureau of Standards, Washington, DC, 1986. 2. Rossini, F.D. et al. (eds), Selected Values of Chemical Thermodynamic Properties, National Bureau of Standards Circular 500, 1952. 3. Tewarson, A., Generation of heat and chemical compounds in fires, in The SFPE Handbook of Fire Protection Engineering, 2nd edn., (eds P.J. DiNenno et al.), Section 3, Chapter 4, National Fire Protection Association, Quincy, Massachusetts, 1995, pp. 3-53 to 3-124. 4. Huggett, C., Estimation of rate of heat release by means of oxygen consumption, J. Fire and Flammability, 1980, 12, 61–5. 5. Tewarson, A., Jiang, F. H. and Morikawa, T., Ventilation-controlled combustion of polymers, Combustion and Flame, 1993, 95, 151–69.

Problems 2.1

Calculate the vapor densities (kg/m3) of pure carbon dioxide (CO2), propane (C3H8) and butane (C4H10) at 25 C and pressure of 1 atm. Assume ideal gas behavior.

2.2

Assuming ideal gas behavior, what will be the final volume if 1 m3 of air is heated from 20 to 700 C at constant pressure?

2.3

Identify the following thermodynamic properties for liquid mixtures: (a) Thermal equilibrium between two systems is given by the equality of ____________. (b) Mechanical equilibrium between two systems is given by the equality of ___________. (c) Phase equilibrium between two systems of the same substance is given by the equality of __________________.

2.4

Show that the pressure rise (p–po) in a closed rigid vessel of volume (V) for the net constant heat addition rate (Q_ ) is

Q_ t p po ¼ o Vcv To po where o is the initial density, po is the initial pressure, To is the initial temperature, t is time and cv is the specific heat at constant volume. The fluid in the vessel is a perfect gas with constant specific heats. 2.5

Formaldehyde (CH2O) burns to completeness in air. Compute: (a) stoichiometric air to fuel mass ratio; (b) mole fraction of fuel in the reactant mixture for an equivalence ratio () of 2; (c) mole fraction of fuel in the product mixture for ¼ 2.

2.6

Hydrogen reacts with oxygen. Write the balanced stoichiometric chemical equation for the complete reaction.

42

THERMOCHEMISTRY

2.7

The mole fraction of argon in a gas mixture with air is 0.1. The mixture is at a pressure of 1000 Pa and 25 C. What is the partial pressure of the argon?

2.8

Compute the enthalpy of formation of propane at 25 C from its chemical reaction with oxygen and its ideal heat of combustion given in Table 2.3.

2.9

Consider the complete stoichiometric reaction for the oxidation of butane:

C4 H10 þ O2 ! CO2 þ H2 O Calculate the heat of formation using data in Tables 2.2 and 2.3. 2.10

Compute the heat transferred in the oxidation of 1 mole of butane to carbon monoxide and water, with reactants and products at 25 C. Use Table 2.2.

nC5 H12 þ 1

1 O2 ! 5CO þ 6 H2 O 2

2.11

The products for the partial combustion of butane in oxygen were found to contain CO2 and CO in the ratio of 4:1. What is the actual heat released per mole of butane burned if the only other product is H2O? The reactants are at 25 C and the products achieve 1000 C. Use average estimates for specific heats. For butane use 320 J/K mole.

2.12

Consider heptane burned in air with reactants and products at 25 C. Compute the heat release per gram of oxygen consumed for (a) CO2 formed and (b) only CO formed. The other product of combustion is water vapor.

2.13

(a) Write the balanced chemical equation for stoichiometric combustion of benzene (C6H6). Assume complete combustion. Calculate the mass of air required to burn a unit mass of combustible. (b) For a benzene–air equivalence ratio of 0.75, write the balanced chemical equation and calculate the adiabatic flame temperature in air. The initial temperature is 298 K and pressure is 1 atm. Assume complete combustion. (c) Calculate the mole fraction for each product of combustion in part (b). (d) Benzene often burns incompletely. If 20 % of the carbon in the benzene is converted to solid carbon and 5 % is being converted to CO during combustion, the remainder being converted to CO2, calculate the heat released per gram of oxygen consumed. How does this compare with the value in Table 2.3?

2.14

(a) For constant pressure processes show that the net heat released for reactions I and II in succession is the same as for reaction III. The initial and final temperatures are 25 C. Use data in the text. I.

CðsÞ þ 12 O2 ðgÞ ! COðgÞ

II.

COðgÞ þ 12 O2 ðgÞ ! CO2 ðgÞ

III.

CðsÞ þ O2 ðgÞ ! CO2 ðgÞ

(b) In reaction II, for oxygen in the reactants: (i) What is the mole fraction of O2?

PROBLEMS

43

(ii) What is the mass fraction of O2? (iii) What is the partial pressure of O2 if the system pressure is 2 atm. 2.15

Carbon monoxide burns in air completely to CO2. (a) Write the balanced chemical equation for this reaction. (b) Calculate the stoichiometric fuel to air ratio by mass. (c) For this reaction, at a constant pressure of 1 atm, with the initial and final temperatures of 25 C, calculate the change in enthalpy per unit mass of CO for a stoichiometric fuel–air mixture. (d) What is the quantity in (c) called? (e) If the final temperature of this reaction is 500 C instead, determine the heat lost from the system per unit mass of CO. (f) Calculate the equivalence ratio for a mixture of five moles of CO and seven moles of air. What species would we expect to find in the products other than CO2?

2.16

Determine the heat of combustion of toluene and express it in proper thermodynamic form. The heat of formation is 11.95 kcal/mole.

2.17

Hydrogen stoichiometrically burns in oxygen to form water vapor. Let cp;i ¼ 1:5 J/g K for all species. (a) For a reaction at 25 C, the enthalpy change of the system per unit mass of H2 is called ______________________. (b) For a reaction at 25 C, the enthalpy change of the system per unit mass of H2O is called ____________________. (c) Compute the enthalpy change of the system per unit mass of water vapor at 1000 C. (d) The initial temperature is 25 C and the temperature after the reaction is 1000 C. Compute the heat transfer for the system per unit mass of H2 burned. Is this lost or added? (e) Repeat (d) for an initial temperature of 300 C.

2.18

Compute the heat of combustion per gram mole of acetonitrile. Acetonitrile (C2H3N(g)) burns to form hydrogen cyanide (HCN(g)), carbon dioxide and water vapor. Heat of formation in kcal=gmol Hydrogen cyanide: Acetonitrile: Water vapor: Carbon dioxide: Oxygen:

2.19

32.3 21.0 57.8 94.1 0.0

An experimentalist assumes that, when wood burns, it can be approximated by producing gaseous fuel in the form of formaldehyde and char in the form of carbon.

44

THERMOCHEMISTRY Heat of formation in kcal=gmol Formaldehyde (CH2O): Carbon (C): Oxygen (O2): Nitrogen (N2): Water vapor (H2O): Carbon dioxide (CO2):

27.7 0 0 0 57.8 94.1

Assume complete combustion in the following: (a) Compute the heat of combustion of CH2O in kJ/kg. (b) Compute the heat of combustion of C in kJ/kg. (c) What heat of combustion would an engineer use to estimate the energy release of wood during flaming combustion? (d) The char yield of wood is 0.1 g C/g wood. Compute the heat of combustion of the solid wood after all of the char completely oxidizes, in kJ/g of wood. 2.20

Determine by calculation the enthalpy of formation in kJ/mole of CH4 given that its heat of combustion is 50.0 kJ/g at 25 C. The heat of formation for carbon dioxide is –394 kJ/mole and water vapor is –242 kJ/mole.

2.21

Nylon (C6H11NO) burns as a gas in air. There is 8 times the amount of stoichiometric air available by moles. The reaction takes place at constant pressure. The initial fuel temperature is 300 C and the air is at 25 C. The reaction produces CO2, H2O (gas) and HCN in the molar ratio, HCN=CO2 ¼ 1=5. Other properties: Heat of formation of nylon ¼ 135 kJ/mole Species Nylon (gas) H2O CO2 HCN N2 O2

Molar specific heats at constant pressure ðJ=K moleÞ 136 50 60 90 35 35

(a) Compute the heat of combustion for this reaction using data in Table 2.2. (b) Compute the temperature of the final state of the products for an adiabatic process. 2.22

Nylon burns to form carbon dioxide, water vapor and hydrogen cyanide, as shown in the stoichiometric equation below:

C6 H11 NO2 þ 6:5 O2 ! 5 CO2 þ 5 H2 O þ HCN The heat of combustion of nylon is 30.8 kJ/g. Find the heat of formation of the nylon in kJ/mole. 2.23

Neoprene (C4H5Cl) is burned in air at constant pressure with the reactants and products at 25 C. The mass yields of some species are found by measurement in terms of g/gF. These product yields are CO at 0.1, soot (taken as pure carbon, C) at 0.1 and gaseous unburned hydrocarbons (represented as benzene, C6H6) at 0.03. The remaining products are water as a gas, carbon dioxide and gaseous hydrogen chloride (HCl).

PROBLEMS

45

(a) Write the chemical equation with molar stoichiometric coefficients, taking air as 1 mole of O2 þ 3:77 moles of N2 (i.e. 4.77 moles of air). (b) Calculate the heat of formation of the neoprene (in kJ/mole) if it is known that the heat of combustion for reaction (a) is 11 kJ/g neoprene. (c) If the reaction were complete to only stable gaseous products, namely water vapor, carbon dioxide and chlorine (Cl2), compute the heat of combustion in kJ/g. (d) For both reactions (a) and (c), compute the energy release per unit mass of oxygen consumed, i.e. [H(reactants)H(products)]/mass of oxygen. Atomic weights: C,12; H,1; O,16; N,14; Cl,35.5, and use Table 2.2. 2.24

Formaldehyde is stoichiometrically burned at constant pressure with oxygen (gas) to completion. CO2 and H2O, condensed as a liquid, are the sole products. The initial temperature before the reaction is 50 C and the final temperature after its completion is 600 C. Find, per mole of fuel, the heat transferred in the process, and state whether it is added or lost. Assume a constant specific heat of 35 J/mole K for all the species. Use Table 2.2 for all of your data.

2.25

The heat of combustion for the reaction times the mass of fuel reacted is equal to: (a) the chemical energy released due to realignment of the molecular structures; (b) the heat lost when the reactants and the products are at 25 C; (c) the chemical energy when the fuel concentration in the reactants is at the upper flammable limit.

2.26

Cane sugar, C12H22O11, reacts with oxygen to form char, gaseous water and carbon dioxide. Two moles of char are formed per mole of sugar cane. Assume the char is pure carbon (graphite). Compute the heat of combustion on a mass basis (kJ/g) for the sugar cane in this reaction. The heat of formation of the sugar cane is 2220 kJ/mole. Cite any data taken from Tables 2.1 and 2.2.

2.27

Hydrogen gas (H2) reacts completely with air in a combustor to form water vapor with no excess air. The combustor is at steady state. If the flow rate of the hydrogen is 1 kg/s, what is the heat transfer rate to or from the combustor? The heat of combustion of hydrogen gas is 121 kJ/g, the initial temperature of the air and hydrogen is 600 K and the exiting temperature of the gases leaving the combustor is 3000 K. Assume constant and equal specific heats of constant pressure, 1.25 kJ/kg K. Show all work, diagrams and reasoning, and indicate whether the heat is from or to the combustor wall.

2.28

Polystryrene can be represented as C8H8. Its heat of combustion for a complete reaction to carbon dioxide and water vapor is 4077 kJ/mole. (a) Compute the heat of formation of C8H8 in kJ/mole. (b) Compute the heat of combustion of C8H8 in kJ/g for the incomplete reaction in air with water vapor in the products as shown below:

17 17 C8 H8 þ ½O2 þ 3:77 N2 ! 6 CO2 þ CO þ C þ 4 H2 O þ ð3:77Þ N2 2 2 2.29

Polystyrene (C8H8)n reacts with the oxygen in air; 4.77 moles of air can be represented as O2 þ 3:77 N2 .

46

THERMOCHEMISTRY (1) 1 C8 H8 þ ðO2 þ3:77 N2 Þ ! CO2 þ H2 OðgÞþ N2 (2) 1 C8 H8 þ ðO2 þ3:77 N2 Þ! CO2 þ H2 OðgÞþ N2 þ COþ C For the incomplete reaction, (2), the molar ratios of the incomplete products are given as CO/ CO2 ¼ 0.07 and C/ CO2 ¼ 0.10 (a) Balance each chemical equation. (b) Compute the heat of combustion for each, on a mass basis. (c) Compute the heat of combustion per unit mass of oxygen used for each reaction. (d) Compute the adiabatic flame temperature for each. (e) Compute the adiabatic flame temperature for each reaction in pure oxygen. Use the information in the chapter 2 to work the problem. You can assume constant specific heats of the species at an appropriate mean temperature (use 1600 K). Also, the heat of formation of polystyrene is – 38.4 kJ/mole.

2.30

(a) Determine the heat of combustion of ethylene, C2H4, for a complete reaction to carbon dioxide and water vapor. (b) Determine its adiabatic flame temperature for a complete reaction with pure oxygen. Assume again water vapor in the products. Use information in Tables 2.1 and 2.2.

2.31

Calculate the adiabatic flame temperatures for the following mixtures initially at 25 C: (a) stoichiometric butane–oxygen mixture; (b) stoichiometric butane–air mixture; (c) 1.8 % butane in air. Use specific heat values of 250 J/K mole for butane and 36 J/K mole for oxygen and nitrogen.

2.32

Calculate the adiabatic flame temperature (at constant pressure) for ethane C2H6 in air: (a) at the lower flammability limit (XL ¼ 3:0%) and (b) at the stoichiometric mixture condition. Assume constant specific heats for all species (1.1 kJ/kg K) and use the data from Tables 2.3 and 4.5. The initial state is 1 atm, 25 C and the initial mixture density is 1.2 kg/m3.

2.33

A gaseous mixture of 2 % (by volume) acetone and 4 % ethanol in air is at 25 C and a pressure of 1 atm. Data Acetone (C3 H6 O) has hc ¼ 1786 kJ/g mol Ethanol (C2 H5 OH) has hc ¼ 1232 kJ/g mol Atomic weights: H ¼ 1 mole, C ¼ 12 moles, O ¼ 16 moles and N ¼ 14 moles Specific heats, cp;i ¼ 1 kJ/kg K, constant for each species (a) For a constant pressure reaction, calculate the partial pressure of the oxygen in the product mixture.

PROBLEMS

47

(b) Determine the adiabatic flame temperature of this mixture. (c) If this mixture was initially at 400 C, what will the resultant adiabatic flame temperature be? 2.34

Polyacrylonitrile (C3H3N) burns to form vapor, carbon dioxide and nitrogen. The heat of formation of the polyacrylonitrile is þ15:85 kcal/g mol (1 cal¼ 4:186 kJ). Use data from Tables 2.1 and 2.2; use specific heat values at 1000 K. (a) Write the balanced chemical equation for the stoichiometric combustion in oxygen. (b) Determine the heat of combustion of the polyacrylonitrile. (c) Write the balanced chemical equation for the stoichiometric combustion in air. (d) Determine the adiabatic flame temperature if the fuel burns stoichiometrically in air.

2.35

Toluene (C7H8) as a gas burns stoichiometrically in air to completion (i.e. forming carbon dioxide and water vapor). Data Heat of formation of toluene is þ11.95 kcal/g mol Species Oxygen Water (gas) Carbon dioxide

Specific heat ðcal=g mol KÞ at 1500 K 8.74 11.1 14.0

(a) For products and reactants at 25 C, compute the heat loss per unit gram mole of toluene consumed. (b) What is the quantity in (a) called? (c) For reactants at 25 C, compute the adiabatic flame temperature. 2.36

Check all correct answers below: (a) For reacting systems, the conservation of mass implies: the preservation of atoms, the conservation of moles, the conservation of molecules, the conservation of species, the equality of mass for the reactants and products in a closed system. (b) Stoichiometric coefficients represent: the number of molecules for each species in the reaction, the number of moles for each species in the reaction, the number of atoms for each species in the reaction, the number of grams for each species in the reaction, the number before the chemical species formulas in a balanced chemical equation.

3 Conservation Laws for Control Volumes

3.1 Introduction A control volume is a volume specified in transacting the solution to a problem typically involving the transfer of matter across the volume’s surface. In the study of thermodynamics it is often referred to as an open system, and is essential to the solution of problems in fluid mechanics. Since the conservation laws of physics are defined for (fixed mass) systems, we need a way to transform these expressions to the domain of the control volume. A system has a fixed mass whereas the mass within a control volume can change with time. Control volumes can address transient problems, allowing us to derive ordinary differential equations that govern uniform property domains. This approach leads to a set of equations that govern the behavior of room fires. This is commonly called a zone model. A control volume can also consist of an infinitesimal or differential volume from which the general partial differential conservation equations can be derived. The solution to such equations falls within the subject of computational fluid mechanics or computational fluid dynamics (CFD) modeling. We will not address the latter. Instead we will use control volume analysis to fit the physics of the problem, and derive the appropriate equations needed. This should give to the student more insight into what is happening and what is relevant. We will take a general and mathematical approach in deriving the conservation laws for control volumes. Some texts adopt a different strategy and the student might benefit from seeing alternative approaches. However, once derived, the student should use the control volume conservation laws as a tool in problem solving. To do so requires a clear understanding of the terms in the equations. This chapter is intended as a reference for the application of the control volume equations, and serves as an extension of thermochemistry to open systems.

Fundamentals of Fire Phenomena James G. Quintiere # 2006 John Wiley & Sons, Ltd ISBN: 0-470-09113-4

50

CONSERVATION LAWS FOR CONTROL VOLUMES

3.2 The Reynolds Transport Theorem The Reynolds transport theorem is a general expression that provides the mathematical transformation from a system to a control volume. It is a mathematical expression that generally holds for continuous and integrable functions. We seek to examine how a function f(x, y, z, t), defined in space over x, y, z and in time t, and integrated over a volume, V, can vary over time. Specifically, we wish to examine d dt

Z Z Z

f ðx; y; z; tÞ dV

ð3:1Þ

VðtÞ

where the volume, V, can vary with time. Since we integrate over V before taking the derivative in time, the integral and therefore the result is only a function of time. A word of caution should be raised. The formidable process of conducting such spatial integration should be viewed as principally symbolic. Although the mathematical operations will be valid, it is rare that in our application they will be so complicated. Indeed, in most cases, these operations will be very simple. For example, if f ðx; y; z; tÞ is uniform in space, then we simply have, for Equation (3.1), dðfVÞ df dV ¼V þf dt dt dt Let us examine this more generally for f ðx; y; z; tÞ and a specified moving volume, VðtÞ. The moving volume can be described in terms of surface, S, to which each point on it is given a velocity, v. That velocity is also defined over the spatial coordinates and time v(x, y, z, t). To complete the description of V in time, we need to not just describe its surface velocity but also the orientation of the surface at each point. This is done by defining a unit vector n which is always oriented outward from the volume and normal to the surface. This is illustrated in Figure 3.1, where S represents a small surface element. As we shrink to the point ðx; y; zÞ, we represent S as a differential element, dS.

Figure 3.1

A description for a moving volume, V

THE REYNOLDS TRANSPORT THEOREM

Figure 3.2

51

Change in volume V over t

Let us consider what happens when V moves and we perform the operations in Equation (3.1). This is depicted in Figure 3.2. Here we have displayed the drawing in two dimensions for clarity, but our results will apply to the three-dimensional volume. In Figure 3.2 the volume is shown for time t, and a small time increment later, t, as a region enclosed by the dashed lines. The shaded regions VI and VII are the volumes gained and lost respectively. For these volumes, we can represent Equation (3.1) by its formal definition of the calculus as d dt

ZZZ VðtÞ

¼ lim

f ðx; y; z; tÞdV

RRR VðtþtÞ

f ðx; y; z; t þ tÞdV

RRR VðtÞ

f ðx; y; z; tÞdV

t

t!0

ð3:2Þ

From Figure 3.2, the first integral on the right-hand side of Equation (3.2) can be expressed as ZZZ ZZZ f ðx; y; z; t þ tÞdV ¼ f ðx; y; z; t þ tÞdV VðtþtÞ

VðtÞ

þ

Z Z Z

f ðx; y; z; t þ tÞdV

V I ðtþtÞ

Z Z Z

f ðx; y; z; t þ tÞdV

ð3:3Þ

V II ðtþtÞ

For volumes VI and VII we consider the integrals as sums over small volumes, Vj , defined by the motion of their corresponding surface elements, Sj . Figure 3.3 provides a (two-dimensional) geometric representation of this process in which Sj moves over S

52

CONSERVATION LAWS FOR CONTROL VOLUMES

Figure 3.3

Relationship between VI and S

to define an intercepted volume VIj . Temporarily, we drop the j-subscript. The motion of S is r in the direction of v where r ¼ v t

ð3:4Þ

with v the magnitude of the velocity vector. By geometry, the volume of a rectilinear region is the product of its cross-sectional area perpendicular to its length times its length. Since two angles are equal whose legs are mutually perpendicular, with this angle being in Figure 3.3, VI ¼ ðSÞðcos Þr

ð3:5Þ

From the definition of the vector dot product, Equations (3.4) and (3.5) can be combined to give v VI ¼ ðSÞ n ðv tÞ ¼ ðv nÞðSÞðtÞ ð3:6Þ v Since v is always more than 90 out of phase with n for VII, the minus sign in Equation (3.3) for the integral is directly accounted for by the application of Equation (3.6) to VII as well. It then follows that ZZZ Z ZZ f ðx; y; z; t þ tÞdV f ðx; y; z; t þ tÞdV VI

¼

N X

VII

f ðx; y; z; t þ tÞv nSj t

ð3:7Þ

j¼1

Combining Equation (3.7) with Equation (3.3), Equation (3.2) can be written as RRR ½ f ðx; y; z; t þ tÞ f ðx; y; z; tÞ dV ZZZ d VðtÞ f ðx; y; z; tÞdV ¼ lim t!0 dt t VðtÞ

PN þ lim

t!0

j¼1

f ðx; y; z; t þ tÞv nSj t t

ð3:8Þ

53

RELATIONSHIP BETWEEN A CONTROL VOLUME AND SYSTEM VOLUME

From the calculus, the limit process formally gives d dt

ZZZ

f ðx; y; z; tÞdV ¼

VðtÞ

Z ZZ ZZZ @f f ðx; y; z; tÞv ndS dV þ @t x;y;z VðtÞ

ð3:9Þ

SðtÞ

where SðtÞ is the surface enclosing the volume VðtÞ. Equation (3.9) is the Reynolds transport theorem. It displays how the operation of a time derivative over an integral whose limits of integration depend on time can be distributed over the integral and the limits of integration, i.e. the surface, S. The result may appear to be an abstract mathematical operation, but we shall use it to obtain our control volume relations.

3.3 Relationship between a Control Volume and System Volume Suppose we now assign a physical meaning to the velocity v, representing it as the velocity of matter in the volume, V. Then if V always contains the same mass, it is a system volume. The properties defined for each point of the system represent those of a continuum in which the macroscopic character of the system is retained as we shrink to a point. Properties at a molecular or atomic level do not exist in this continuum context. Furthermore, since the system volume is fixed in mass, we can regard volume V to always enclose the same particles of matter as it moves in space. Each particle retains its continuum character and thermodynamic properties apply. Let us see how to represent changes in properties for a system volume to property changes for a control volume. Select a control volume (CV) to be identical to volume VðtÞ at time t, but to have a different velocity on its surface. Call this velocity, w. Hence, the volume will move to a different location from the system volume at a later time. For example, for fluid flow in a pipe, the control volume can be selected as stationary ðw ¼ 0Þ between locations 1 and 2 (shown in Figure 3.4, but the system moves to a new location later in time. Let us apply the Reynolds transport theorem, Equation (3.9), twice: once to a system volume, VðtÞ, and second to a control volume, CV, where CV and V are identical at time t. Since Equation (3.9) holds for any well-defined volume and surface velocity distribution, we can write for the system d dt

ZZZ VðtÞ

Figure 3.4

f dV ¼

ZZZ VðtÞ

ZZ @f f v n dS dV þ @t SðtÞ

Illustration of a fixed control volume for pipe flow and a system

ð3:10Þ

54

CONSERVATION LAWS FOR CONTROL VOLUMES

and for the control volume d dt

ZZZ

f dV ¼

CV

ZZ

ZZ @f f w n dS dV þ @t

CV

ð3:11Þ

CS

The first terms on the right-hand side of Equations (3.10) and (3.11) are identical since CV ¼ VðtÞ at time t. Furthermore, since SðtÞ ¼ CSðcontrol surfaceÞ at time t, on substituting Equation (3.11) from Equation (3.10) and rearranging, we obtain d dt

ZZZ

f dV ¼

d dt

ZZZ CV

VðtÞ

f dV þ

ZZ

f ðv wÞ n dS

(3.12)

CS

This is the relationship we seek. It says that the rate of change of a property f defined over a system volume is equal to the rate of change for the control volume plus a correction for matter that carries f in or out. This follows since v w is the relative velocity of matter on the boundary of the control volume. If v w ¼ 0, no matter crosses the boundary. As we proceed in applying Equation (3.12) to the conservation laws and in identifying a specific property for f , we will bring more meaning to the process. We will consider the system to be composed of a fluid, but we need not be so restrictive, since our analysis will apply to all forms of matter.

3.4 Conservation of Mass In general, we consider a fluid system that consists of a reacting mixture. Let , the density of the mixture, be identified for f. Then from Equation, (2.11b) and (3.12), the conservation of mass for a control volume is d dt

ZZZ CV

dV þ

ZZ

ðv wÞ n dS ¼ 0

(3.13)

CS

The surface integral represents the net flow rate of mass out of the control volume. This is easily seen from Figure 3.5, where v w is the relative velocity at the fluid giving a flow

Figure 3.5

Flow rate from a control volume

55

CONSERVATION OF MASS

of fluid particles across S. Only the component of the particle motion along n can cross the boundary. We have already seen that v w nðSÞðtÞ is the volume of fluid crossing S in time t by Equation (3.6). It follows that m_ ¼ ðv wÞ n S

ð3:14Þ

is the mass flow rate of the fluid through S. Equation (3.13) can be expressed in words as

Rate of change for sum of all mass sum of all mass þ ¼0 mass within CV flow rates out flow rates in

An alternative form to Equation (3.13) for discrete regions of mass exchange, j, is ! ! X X dm þ m_ j m_ j ¼ 0 (3.15) dt CV j¼1 j¼1 out

in

where the sum is implied to be over all flow paths from and to the control volume. Example 3.1 A balloon filled with air at density allows air to escape through an opening of cross-sectional area, S0 , with a uniform velocity, v0 , given by a measurement device on the balloon. Derive an expression for the volume VB of the balloon over time. Solution Select the control volume to always coincide with the surface of the balloon and to be coincident with the plane of the exit velocity as shown in Figure 3.6. From Equation (3.13), d dt

ZZZ

Figure 3.6

dV ¼

d VB dt

Shrinking balloon problem

56

CONSERVATION LAWS FOR CONTROL VOLUMES

regarding as constant for air, and ZZ ðv wÞ n dS ¼ v0 S0 since only the exit surface contributes with v w on the remainder of the control surface. The density and the exit term are constant, so by integration with VB;o , the initial volume, VB ¼ VB;0 v0 S0 t

3.5 Conservation of Mass for a Reacting Species To consider the control volume form of the conservation of mass for a species in a reacting mixture volume, we apply Equation (2.14) for the system and make the conversion from Equation (3.12). Here we select f ¼ i , the species density. In applying Equation (3.13), v must be the velocity of the species. However, in a mixture, species can move by the process of diffusion even though the bulk of the mixture might be at rest. This requires a more careful distinction between the velocity of the bulk mixture and its individual components. Indeed, the velocity v given in Equation (3.13) is for the bulk mixture. Diffusion velocities, V i , are defined as relative to this bulk mixture velocity v. Then, the absolute velocity of species i is given as vi ¼ v þ V i

ð3:16aÞ

with the bulk velocity defined as v

N X

i vi

ð3:16bÞ

i¼1

where N is the number of species in the mixture. The mass flow rate per unit area, or species mass flux, due to diffusion is given by Fick’s law: i V i ¼ Di rYi

ð3:17Þ

where Di is a diffusion coefficient for species i in the mixture. This expression of mass transfer of a species is analogous to Fourier’s law for the heat flux due to conduction. In order always to include species i in the system before converting to a control volume, v must be selected as vi . Then we obtain from Equations (2.14) and (3.12), d dt

ZZZ CV ZZ

þ

CS

i dV þ

ZZ

i ðv wÞ n dS

CS

i V i ndS ¼ m_ i;r

(3.18)

CONSERVATION OF MASS FOR A REACTING SPECIES

57

In words, this reads as 3 2 3 3 2 3 2 rate of pronet mass net bulk Rate of 7 6 6 change 7 6 mass flow 7 6 flow rate 7 6 duction of 7 7 6 mass for 7 7 6 7 6 6 7 6 of mass 7 6 rate of 7 6 of species i 7 6 7 7 6 7 6 7þ6 6 7 6 species i 7 þ 6 due to def- 7 ¼ 6 species i due 7 6 for 7 6 to chemical 7 7 6 7 6 6 7 4 species i in 5 4 from the 5 4 fusion from 5 6 4 reaction within 5 the CV CV the CV the CV 2

Since the chemical reaction may not necessarily be uniformly dispersed over the control volume, it is sometimes useful to represent it in terms of a local reaction rate, m_ 000 i;r , or the mass production rate of species i per unit volume. Then ZZZ m_ i;r ¼ ð3:19Þ m_ 000 i;r dV CV

Usually for applications to combustors or room fires, diffusion effects can be ignored at surfaces where transport of fluid occurs, but within diffusion flames these effects are at the heart of its transport mechanism. A more useful form of Equation (3.18) is X X dmj þ ð3:20Þ m_ i; j þ m_ i;d; j ¼ m_ i; r dt CV j; net out j; net out where the sums are over all j flow paths; m_ i;j represents the bulk flow rate of species i and m_ i;d;j represents the diffusion flow rate at each flow path. The diffusion mass flux is given by Equation (3.17). Where gradients of the species concentration ðYi Þ are zero across a flow path, then m_ i;d is zero. When concentrations are uniform throughout the control and uniform at entering flow paths, Equation (3.20) can be expressed in terms of Yi . The mass of species in the control volume can be given as mi ¼ m Yi

ð3:21Þ

where m is the mass in the CV and Yi is its mass fraction. At entering j flow paths the mass fraction for each species i is represented as Yi;j . Then Equation (3.20) becomes ! ! X X X dðYi mÞ þ m_ j Yi m_ j Yi; j þ m_ i;d; j ¼ m_ i; r ð3:22Þ dt j j j; net out out

in

By multiplying Equation (3.15) by Yi and then subtracting it from Equation (3.22), we obtain for this uniform concentration case

m

X X dYi þ m_ j ðYi Yi; j Þ þ m_ i;d; j ¼ m_ i; r dt j; in only j; net out

(3.23)

58

CONSERVATION LAWS FOR CONTROL VOLUMES

Figure 3.7

Combustor problem

Note that the bulk out-flow contribution has cancelled. Example 3.2 Consider a combustor using air and methane under steady conditions. A stoichiometric reaction is assumed and diffusion effects are to be neglected. The reactants and products are uniform in properties across their flow streams. Find the mass fraction of carbon dioxide in the product stream leaving the combustor. Solution The combustor is shown in Figure 3.7 with the control volume. The stoictiometric reaction is CH4 þ 2ðO2 þ 3:76 N2 Þ ! CO2 þ 2 H2 O þ 7:52 N2 Since the conservation equations are in terms of mass, it is useful to convert the stoichiometric coefficients to mass instead of moles. The condition of steady state means that all properties within the control volume are independent of time. Even if the spatial distribution of properties varies within the control volume, by the steady state condition the time derivative terms of Equations (3.13) and (3.18) are zero since the control volume is also not changing in size. Only when the resultant integral is independent of time can we ignore these time derivative terms. As a consequence, the conservation of mass becomes m_ P m_ CH4 m_ A ¼ 0 and the conservation of CO2 becomes m_ P YCO2 ¼ m_ CO2 ;r Since we have a stoichiometric reaction, all of the entering methane is consumed: m_ CH4 ¼ ðm_ CH4 ;r Þ Recall that in our convention m_ i;r is positive in sign when it is produced. From Example 2.2, the molecular weight of air is 28.84 g/mole air. By the chemical equation m_ A ¼

ð2Þð4:76 moles airÞð28:84 g=mole airÞ m_ CH4;r ¼ 17:16 m_ CH4 ð1mole CH4 Þð16 g=mole CH4 Þ

CONSERVATION OF MOMENTUM

59

Also ð1mole CO2 Þð44 g=mole CO2 Þ ðm_ CH4 ;r Þ ð1mole CH4 Þð16 g=mole CH4 Þ ¼ ð2:75Þðm_ CH4 ;r Þ

m_ CO2;r ¼

Combining m_ CO2 ;r m_ P ð2:75Þðm_ CH4 ;r Þ ¼ m_ A þ m_ CH4 ð2:75Þm_ CH4 ¼ ð18:16Þm_ CH4

YCO2 ¼

YCO2 ¼ 0:151

3.6 Conservation of Momentum The conservation of momentum or Newton’s second law applies to a particle or fixed set of particles, namely a system. The velocity used must always be defined relative to a fixed or inertial reference plane. The Earth is a sufficient inertial reference. Therefore, any control volume associated with accelerating aircraft or rockets must account for any differences associated with how the velocities are measured or described. We will not dwell on these differences, since we will not consider such noninertial applications. The law in terms of a control volume easily follows from Equations (2.12) and (3.12) for the bulk system with f selected as vx , taking vx as the component of v in the x direction: X

Fx ¼

d dt

ZZZ CV

vx dV þ

ZZ

vx ðv wÞ n dS

(3.24)

CS

In words, 3 2 3 3 2 rate of change net rate of Net sum of all 6 forces acting on 7 6 of momentum in 7 6 momentum in 7 6 7 6 7 6 7 6 the fluid in the 7 ¼ 6 the x direc- 7 þ 6 the x direc- 7 7 6 7 6 7 6 4 CV in the x 5 4 tion for the 5 4 tion flowing 5 direction fluid in the CV out of the CV 2

Example 3.3 Determine the force needed to hold a balloon filled with hot air in place. Consider the balloon to have the same exhaust conditions as in Example 3.1 and the hot air density, , is constant. The cold air density surrounding the balloon is designated as 1 . Assume the membrane of the balloon to have zero mass.

60

CONSERVATION LAWS FOR CONTROL VOLUMES

Figure 3.8

Hot balloon held in place by force F

Solution Select a control volume as in Example 3.1 (see Figure 3.8). Let us examine the right-hand side of Equation (3.24) using the sign convention that vertically up is positive. First, we approximate ZZZ d vx dV 0 dt because the velocity of the air in the balloon is stationary except near the shrinking membrane and at the exit plane. If the membrane had a significant mass we might need to reconsider this approximation. However, if the balloon were to retain a spherical shape while shrinking, the spherical symmetry would nearly cancel the vertical momentum between the upper and lower hemispheres. Since we expect v0 to be more significant, the second term becomes ZZ vx ðv wÞ n dS ¼ ðv0 Þðþv0 Þ S0 Let us now consider the sum of the forces upwards. F is designated as the force, downwards, needed to hold the balloon in place. Another force is due to the pressure distribution of the cold atmosphere on the balloon. Assuming the atmosphere is motionless, fluid statistics gives the pressure variation with height x as @p ¼ 1 g @x The pressure acts normal everywhere on the membrane, but opposite to n. This force can therefore be expressed as ZZ ðpnÞ dS i CS

CONSERVATION OF ENERGY FOR A CONTROL VOLUME

61

where i is the unit vector upwards. We can struggle with this surface integration or we can resort to the divergence theorem, which allows us to express ZZ

ðpiÞ n dS ¼

ZZZ

r ðpiÞ dV ¼

ZZZ

ðrp iÞ dV

CS

since the unit vector i is constant in magnitude, i.e. r i ¼ 0. However, rp i ¼

@p ¼ 1 g @x

Therefore, ZZ

ðpnÞ dS i ¼

CS

ZZZ

ð1 gÞ dV ¼ þ1 g VB

CV

This is a very important result which is significant for problems involving buoyant fluids. Finally, the last force to consider is the weight for the balloon consisting of the massless membrane and the hot air: g VB . Combining all the terms into Equation (3.24) gives F g VB þ 1 g VB ¼ v20 S0 or solving for F, the needed force, F ¼ ð1 Þg VB þ v20 S0 The first term on the right-hand side is known as the buoyant force, the second is known as thrust. If this were just a puff of hot air without the balloon exhaust, we RRR would only have the buoyant force acting. In this case we could not ignore ðd=dtÞ vx dV since the puff would rise ðvx þÞ solely due to its buoyancy, with viscous effects retarding it. Buoyancy generated flow is an important controlling mechanism in many fire problems.

3.7 Conservation of Energy for a Control Volume As before, we ignore kinetic energy and consider Equation (2.18) as a rate equation for _ s , and pressure, W _ p. the system separating out the rate of work due to shear and shafts, W Then

dU dt

_sW _p ¼ Q_ W System

ð3:25Þ

62

CONSERVATION LAWS FOR CONTROL VOLUMES

The internal energy of the system is composed of the contributions from each species. Therefore it is appropriate to represent U as

U¼

N Z Z Z X i¼1

i ui dV

ð3:26Þ

VðtÞ

A subtle point is that the species always occupy the same volume VðtÞ as the mixture. We can interchange the operations of the sum and the integration. Then, taking f as i ui for a system volume always enclosing species i, we need to use vi as our velocity, as was done in Equation (3.16a), to obtain ZZZ X ZZ X N n dU d ¼ i ui dV þ i ui ðv þ Vi wÞ n dS dt System dt i¼1 i¼1 CV

ð3:27Þ

CS

It is sometimes useful to write the rate of heat added to the control volume in terms of a heat flux vector, q_ 00 ; as Q_ ¼

ZZ

q_ 00 n dS

ð3:28Þ

CS

where q_ 00 is taken as directed out of the control volume or system. Note that when we apply the transformation to the Q and W terms there is no change since the system coincides with the control volume at time t. As with the mass flux of species in Equation (3.17), we can express the heat flux for pure conduction as q_ 00 ¼ krT

ð3:29Þ

where k is the thermal conductivity. In addition, q_ 00 would have a radiative component. The pressure work term is also made up of all of the species. However, let us first _ has the consider it for a single species and not include any subscripts. Since W convention of ‘work done by the system on the surroundings’ and the definition of work is the product of force times distance moved in the direction of the force, we can show from Figure 3.9 that _ p ¼ pn vðSÞ W

ð3:30Þ

for element S that is displaced. For all of the species within the control volume, the total pressure work is _p ¼ W

ZZ X N CS

i¼1

pi n vi dS

ð3:31Þ

CONSERVATION OF ENERGY FOR A CONTROL VOLUME

Figure 3.9

63

Work due to pressure on the surroundings

where pi is the partial pressure of species i. Combining Eqations (3.27), (3.28) and (3.31) into Equation (3.25) gives d dt

ZZZ CV

¼

ZZ

N X

! i ui dV þ

i¼1

_s q_ 00 n dS W

CS

ZZ X N CS

i ui ðv þ V i wÞ n dS

i¼1

ZZ X N CS

pi ðv þ V i Þ n dS

Introducing the species enthalpy as hi ¼ ui þ pi =i and rearranging, we obtain ! ZZZ X ZZ X N N d ði hi pi Þ dV þ i hi ðv wÞ n dS dt i¼1 i¼1 CV CS ! ! ZZ X ZZ N N X i hi Vi n dS W s pi w n dS ¼ q_ 00 þ CS

i¼1

ð3:32Þ

i¼1

CS

ð3:33Þ

i¼1

PN By Dalton’s law, Equation (2.9), the mixture pressure, p, is i¼1 pi . The term P N V h is sometimes considered to be a heat flow rate due to the transport of i¼1 i i i enthalpy by the species. (This is not the same as q_ 00 arising from rYi , which is called the Dufour effect and is generally negligible in combustion.) With the exception of the enthalpy P diffusion term, all the sums can be represented in mixture properties since h ¼ Ni¼1 i hi . However, it is convenient to express the enthalpies in terms of the heat of formation and specific heat terms, and then to separate these two parts.

64

CONSERVATION LAWS FOR CONTROL VOLUMES

From Equation (2.24) we substitute for hi. Without a loss in generality, but to improve the clarity of our discussion, we consider a three-component reaction Fuel ðFÞ þ oxygen ðO2 Þ ! products ðPÞ In terms of the molar stoichiometric coefficients F F þ O 2 O 2 ! P P Let us represent the mass consumption rate of the fuel per unit volume (of mixture) to be 000 þm_ F;r . This is opposite to the sign convention adopted in Equation (3.18) in which 000 species produced are positive in sign. However, it is more natural to treat m_ F;r as inherently positive in sign. Then 000

m_ O2 ;r

O2 MO2 000 ¼ m_ F;r F MF

ð3:34aÞ

P MP 000 ¼ m_ F;r F MF

ð3:34bÞ

and 000

m_ P;r

Substituting into Equation (3.33) gives d dt

Z Z Z "X Z 3 i hf;i þ i¼1

CV

þ

ZZ X Z 3 i hf;i þ CS

¼

ZZ

i¼1

_s q_ 00 n dS W

25 C T

25 C

ZZ

CS

T

cpi dT

# þ p dV

cpi dt ðv þ V i wÞ n dS

pw n dS

ð3:35Þ

CS

Since hf;i is a constant for each species, we can multiply it by each term in its respective species conservation equation, given by Equation (3.18). Summing over i ¼ 1 to 3 for all species and using Equations (3.19) and (3.34), we obtain d dt

ZZZ CV

¼

3 X

ZZZ

i¼1

! i hf;i

dV þ

ZZ X 3 CS

i hf;i ðv þ V i wÞ n dS

i¼1

O 2 MO2 P MP 000 m_ F hf;F þ hf;O2 hf;P dV F MF F MF

ð3:36Þ

CV

The quantity in the brackets is exactly the heat combustion in mass terms, hc , which follows from Equation (2.25). Now subtracting Equation (3.36) from Equation (3.35), we

65

CONSERVATION OF ENERGY FOR A CONTROL VOLUME

obtain (for i ¼ 1; N) ! # ZZZ " X Z T N d i cpi dT p dV dt 25 C i¼1 CV ! Z T ZZ X N i cpi dt ðv þ V i wÞ n dS þ ¼

ZZ

CS

25 C

i¼1

q_ 00 n dS þ hc

ZZZ

CS

000 _s m_ F dV W

CV

ZZ

(3.37)

pw n dS

CS

This is a general form of the energy equation for a control volume useful in combustion problems. The terms can be literally expressed as 3 2 3 2 net rate of enthalpy Rate of change for 7 6 the internal energy 7 6 flow due to bulk 7 6 7 6 7 6 7þ6 4 of the mixture in 5 4 flow and diffusion flow 5 from the CV the CV 3 2 3 net rate of rate of chemical 6 7 6 7 ¼ 4 heat added 5 þ 4 energy released 5 2

within the CV 3 2 3 net rate of shaft rate of pressure 6 7 6 7 4 and shear work done 5 4 work due to 5 by the fluid in the CV a moving CV 2

to the CV

A series of approximations will make Equation (3.37) more practical for engineering applications. These approximations are described below, and the resulting transformed equations are presented as each approximation is sequentially applied. 1. Equal specific heats, P cpi ¼ cp ðTÞ. This assumption allows the diffusion term involving i V i to vanish since Ni¼1 i V i ¼ 0 from Equation (3.16). 2. Equal and constant specific heats, cp ¼ constant. From the Reynolds transport theorem, Equation (3.9), the pressure terms can be combined as ZZZ ZZZ ZZ d @p p dV ¼ pw n dS ð3:38Þ dV þ dt @t CV

CV

CS

Also for all specifics heats equal and independent of temperature, i.e. cpi ¼ cp ¼ constant, Equation (3.37) becomes cp

d dt

ZZZ CV

¼ hc

ðT 25 CÞdV ZZZ CV

000

m_ F;r dV

ZZZ

ZZ CS

@p dV þ cp @t

CV

_ shaft q_ 00 n dS W

ZZ

ðT 25 CÞðv wÞn dS

CS

ð3:39Þ

66

CONSERVATION LAWS FOR CONTROL VOLUMES

P where the diffusion term has been eliminated by Equation (3.16a) as i V i ¼ 0. Finally, by multiplying the conservation of mass, Equation (3.13), by 25 C and subtracting from Equation (3.39), it follows that ZZZ ZZ ZZZ d @p dV þ T dV cp Tðv wÞ n dS cp dt @t CV CV CS ZZZ ZZ 000 00 _ shaft ¼ m_ F hc dV q_ n dS W CV

ð3:40Þ

CS

3. Uniform fluxes and properties. For the case of discrete fluxes of mass and heat at the surface of the control volume with uniform boundary temperatures at these regions, the equation can be expressed as d cp dt

ZZZ

T dV

CV

¼

ZZZ

ZZZ CV

X @p dV þ m_ j cp Tj @t j;net;out

000

_ shaft m_ F hc dV þ Q_ added W

ð3:41Þ

CV

For uniform properties within the control volume and discrete j-flows at the control surface cp

X d dp _ shaft ðmTÞ V þ m_ j cp Tj ¼ m_ F;reacted hc þ Q_ W dt dt j;net;out

ð3:42Þ

If the fluid mixture is a perfect gas following pV ¼ RmT

ð3:43Þ

where R the gas constant is given as R ¼ cp cv

ð3:44Þ

we can write

cv

X d dV _ shaft ðmTÞ þ p þ m_ j cp Tj ¼ m_ F;reacted hc þ Q_ W dt dt j;net;out

(3.45)

Alternatively, for a perfect gas, we can use Equation (3.42) in a modified form. The first term on the left can be expanded as

cp

d dT dm ðmTÞ ¼ cp m þ cp T dt dt dt

ð3:46Þ

67

CONSERVATION OF ENERGY FOR A CONTROL VOLUME

Figure 3.10

A control volume encasing an expanding smoke layer in a compartment

The conservation of mass is multiplied by cp T to give cp T

X dm þ m_ j cp T ¼ 0 dt j;net;out

ð3:47Þ

Subtracting Equation (3.47) from Equation (3.42) gives, for a perfect gas mixture with uniform properties in the control volume,

cp m

X dT dp _ shaft V þ m_ j cp ðT Tj Þ ¼ m_ F;reacted hc þ Q_ W dt dt j;in only

(3.48)

Here T is the uniform temperature in the CV. Equations (3.45) and (3.48) are all equivalent under the three approximations, and either could be useful in problems. The development of governing equations for the zone model in compartment fires is based on these approximations. The properties of the smoke layer in a compartment have been described by selecting a control volume around the smoke. The control volume surface at the bottom of the smoke layer moves with the velocity of the fluid there. This is illustrated in Figure 3.10. Example 3.4 A turbulent fire plume is experimentally found to burn with 10 times the required stoichiometric air up to the tip of the flame. It is also measured that 20 % of the chemical energy is radiated to the surroundings from the flame. The fuel is methane, which is supplied at 25 C and burns in air which is also at 25 C. Calculate the average temperature of the gases leaving the flame tip. Assume constant and equal specific heats and steady state. Solution A fixed control volume is selected as a cylindrical shape with its bases at the burner and flame tip respectively, as shown in Figure 3.11. The radius of the CV cylinder is just large enough to contain all of the flow and thermal effects of the plume. This is illustrated in Figure 3.11 with the ambient edge of the temperature and velocity distributions included in the CV. From the conservation of mass, Equation (3.15), m_ m_ A m_ CH4 ¼ 0

68

CONSERVATION LAWS FOR CONTROL VOLUMES

Figure 3.11 Fire plume problem

Apply the conservation of energy, Equation (3.40). Since the control volume is fixed the pressure work term does not apply. The shear work (v shear force) is zero because (a) the radius of the control volume was selected so that the velocity and its gradient are zero on the cylindrical face; and (b) at the base faces, the velocity is normal to any shear surface force. Similarly, no heat is conducted at the cylindrical surface because the radial temperature gradient is zero, and conduction is ignored at the bases since we assume the axial temperature gradients are small. However, heat is lost by radiation as Q_ ¼ ð0:20Þ m_ F;r hc Here m_ F;r is defined as the production rate of fuel. The heat of combustion of the methane is computed from Equation (2.25) or its mass counterpart in Equation (3.36). It can be computed from the stoichiometric reaction involving oxygen, or air, or 10 times stoichiometric air without change since any diluent or species not taking part in the reaction will not contribute to this energy. Therefore, we use CH4 þ 2 O2 ! CO2 þ 2 H2 O and obtain, using Table 2.2,

ð74:9 kJ=mole CH4 Þ 1 mole CO2 44 g=mole CO2 hc ¼ þ0 ð16 g=mole CH4 Þ 1 mole CH 4 16 g=mole CH4 393:5 KJ=mole CO2 2 moles H2 O 18 g=mole H2 O þ 1 mole CH4 16 g=mole CH4 44 g=mole CO2

241:8 KJ=mole H2 O hc ¼ 50:1 kJ=g CH4 18 g=mole H2 O

CONSERVATION OF ENERGY FOR A CONTROL VOLUME

69

Alternatively, we could have taken this value from data in the literature as given by Table 2.3. Since we have 2 moles of stoichiometric oxygen, we must have 2(4.76) moles of stoichiometric air. For a molecular weight of air of 28.84 g/mole air, we compute h i ð2Þð4:76Þmoles of air ð28:84 g=mole of airÞ for every mole of methane consumed. The mass rate of air consumed is ð2Þð4:76Þð28:84Þg of air m_ CH4 ;r ð1 mole CH4 Þð16g=mole CH4 Þ ¼ 17:16 m_ CH4 ;r

ðm_ A Þst ¼

The supply rate of air is 10 times stoichiometric , or m_ A ¼ 171:6 m_ CH4 ;r Since m_ CH4 ;r is equal to the fuel supply rate ðm_ CH4 Þ by definition for a stoichiometric reaction, then the exit mass flow rate is m_ ¼ m_ A þ m_ CH4 ¼ ð171:6 þ 1Þ m_ CH4 ;r The energy equation can now be written from Equation (3.40) as m_ cp T m_ A cp TA m_ CH4 cp TCH4 ¼ ð0:20Þðm_ CH4 ;r Þhc þ m_ CH4 ;r hc We need to select a suitable value for cp. Since air contains mostly nitrogen and so does the heavily diluted exit stream, a value based on nitrogen can be a good choice. From Table 2.1 at 600 K cp ¼

30:1J=mole K ¼ 1:08 J=g K 28 g=mole

For methane at 25 C or 298 K, cp ¼ 2:24 J=g K. Note that on substituting for the temperatures in this steady state example it makes no difference whether K or C units are used. This follows from the conservation of mass. However, for unsteady applications of Equation (3.40), since we have used the perfect gas law in which T is in K, we should be consistent and use it through the equations. When in doubt, use K without error. Substituting: ð171:6 þ 1Þðm_ CH4 ;r Þð1:08 103 kJ=g KÞT 171:6ðCH4 ;r Þð1:08 103 kJ=g KÞ ðCH4 ;r Þð1:08 10

3

kJ=g KÞ

¼ ð0:8ÞðCH4 ;r Þð50:1 kJ=g Þ

ð298 KÞ

ð298 KÞ

70

CONSERVATION LAWS FOR CONTROL VOLUMES

or T¼

ð0:8Þð50:1Þ þ ð172:6Þð1:08 103 Þð298Þ ð172:6Þð1:08 103 Þ

T ¼ 513:0 K ¼ 240 C This is a realistic approximate average temperature at the flame tip for most fire plumes.

Problems 3.1

Water flows into a tank at a uniform velocity of 10 cm/s through a nozzle of cross-sectional area 4 cm2. The density of water is 1 kg/m3. There is a leak at the bottom of the tank that causes 0.5 g/s of water to be lost during filling. The tank is open at the top, has a crosssectional area of 150 cm2 and is 20 cm high. How long does it take for the tank to fill completely? For a control volume enclosing the water in the tank, is the process steady?

3.2

Propane and air are supplied to a combustion chamber so that 20 g/s of propane reacts. The reaction forms H2O, CO2 and CO where the molar ratio of CO to CO2 is 0.1. The exhaust gases flow at a rate of 360 g/s. Assuming the process is steady and conditions are uniform at the exit, compute the exit mass fraction of the CO.

3.3

Methane is supplied to a burner in a room for an experiment at 10 g/s. The measurements indicate that air flow into the room doorway is 800 g/s and the door is the only opening. The exhaust leaves the room through the upper part of the doorway at a uniform temperature of 400 C. Assume that the methane burns completely to CO2 and H2O and that steady conditions prevail in the room. (a) Calculate the rate of heat loss to the surroundings. (b) Calculate the mass fraction of CO2 leaving the room.

3.4

Flashover occurs in a room causing 200 g/s of fuel to be generated. It is empirically known that the ‘yield’ of CO is 0.08. Yield is defined as the mass of product generated per mass of fuel supplied. The room has a volume of 30 m3 and is connected to a closed corridor that has a volume of 200 m3. The temperatures of the room and corridor are assumed uniform and constant at 800 and 80 C respectively. An equation of state for these gases can be taken as T ¼ 360 kg K/m3. Steady mass flow rates prevail at the window of the room (to the outside air) and at the doorway to the corridor. These flow rates are 600 and 900 g/s respectively. Derive expressions for the mass fraction of CO in the room and corridor as a function of time assuming uniform concentrations in each region.

3.5

Carbon monoxide is released in a room and burns in a pure oxygen environment; e.g. perhaps an accident on a space station. The combustion is given by the chemical equation

1 CO þ O2 ! CO2 2 (a) What is the stoichiometric fuel (CO) to oxygen (O2) mass ratio, r, for this reaction? (b) If the flow rate of oxygen supplied to the room is 500 g/s and the release rate of CO is 200 g/s, what is the rate of combustion of the CO?

PROBLEMS

71

(c) If the flow rate of oxygen remains as given in (b) but the release rate of CO suddenly changes to 1000 g/s, what is the mass fraction of CO in the room? (d) For the case (b), what is the mass fraction of CO2 leaving the room?

m· out m· O2 ·

m CO

3.6

C3H8 is burned with 10 times the stoichiometric air in a steady flow process. The reaction is complete, forming CO2 and H2O. The fuel and air are mixed at 400 C before entering the combustor. The combustor is adiabatic. Specific heats are all constant, cp ¼ 1 J/g. (a) Calculate the stoichiometric air to fuel mass ratio. (b) Calculate the stoichiometric oxygen to fuel mass ratio (c) If the flow rate of the C3H8 is 10 g/s, calculate the exit flow rate. (d) Calculate the oxygen mass fraction at the exit. (e) Calculate the C3H8 mass fraction at the inlet. (f) Calculate the exit temperature. (g) Calculate the enthalpy per unit mass for the H2O(g) in the exit stream (with respect to the 25 C reference state).

3.7

An oxygen consumption calorimeter is shown below in which a chair is burned with all of the combustion products captured by the exhaust system. The flow rate of the gases measured in the stack at station A is 2.0 m3/s, and the temperature of the mass fraction and O2 are 327 C and 0.19 respectively. Assume steady state conditions. The surrounding air for combustion is at 20 C. The specific heat of air is 1 kJ/kg K and its density is 1.1 kg/m3. State clearly any other necessary assumptions in working the problem. 2.0 m3/s

A

A

Chair

TA = 327 ˚C YO2,A = 0.19

72

CONSERVATION LAWS FOR CONTROL VOLUMES Compute: (a) the mass flow rate in the exhaust stack; (b) the oxygen consumption rate; (c) the rate of chemical energy released by the chair fire (use the approximation that 13 kJ are released for every gram of O2 consumed); (d) the rate of heat lost from the flame and exhaust gases up to station A.

3.8

A fire occurs in a space station at 200 kW. The walls can be considered adiabatic and of negligible heat capacity. The initial and fuel temperatures are at 25 C. Assume the station atmosphere has uniform properties with constant specific heats as given. Assume that the constant and equal specific heats of constant pressure and volume are 1.2 and 1.0 kJ/kg K respectively. Conduct your analysis for the control volume (CV) consisting of the station uniform atmosphere, excluding all solids and the fuel in its solid state.

125 m3

200 kW fire

Note that p ¼ RT, where p is pressure. (a) Indicate all mass exchanges for the CV shown and their magnitudes. (b) Apply conservation principles to the CV and determine an equation for the atmosphere temperature in the CV. Identify your terms clearly. (c) How can the pressure be determined? Explain. Write an equation. 3.9

At the instant of time shown, the entrainment of air into the fire plume is 300 g/s, the outflow of smoke through the door is 295 g/s and the liquid fuel evaporation rate of a spill burning on the floor is 10 g/s. What is the mass rate of smoke accumulation within the room?

Smoke

3.10

A chair burns in a room, releasing 500 kW, and the heat of combustion for the chair is estimated as 20 kJ/g. Fire and room conditions can be considered steady. The air flow rates into a door and through a window are 200 and 100 g/s respectively. Assume the room is adiabatic. The gases have constant and equal specific heats ¼1.5 kJ/kg K. The ambient air is 25 C and the chair fuel gasifies at 350 C. Compute the average temperature of the room gases. Account for all terms and show all work.

PROBLEMS 3.11

73

A furnace used preheated air to improve its efficiency. Determine the adiabatic flame temperature (in K) when the furnace is operating at a mass air to mass fuel ratio of 16. Air enters at 600 K and the fuel enters at 298 K. Use the following approximate properties: All molecular weights are constant at 29 g/g mol. All specific heats are constant at 1200 J/kg K. Enthalpy of formation: zero for air and all products, 4 107 J/kg for the fuel.

3.12

Methane gas burns with air in a closed but leaky room. The leak in the room ensures that the room pressure stays equal to the surrounding atmosphere, which is at 105 Pa and 20 C. The room is adiabatic and no heat is lost to the boundary surfaces. When the methane burns the flow is out of the room only. Assume that the gas in the room is well mixed. Its initial temperature is 20 C and its volume is 30 m3. The power of the fire during this combustion is Q_ ¼ 30 kW. The specific heat of the mixture is constant and equal to 1 J/g K. (Hint: pV ¼ mRT, where R ¼ (8.314 J/mole K)/(29 g/mole), 1 N/m2 ¼ 1 Pa, 1 N ¼ 1 kg m/s2 and 1 J ¼ 1 N m.) (a) Show that the mass flow rate out is Q_ =cp T, where T is the gas temperature in the room. (b) Compute the temperature in the room after 2 seconds of combustion.

3.13

A room has mass flow rates of fuel, air and fire products at an instant of time in the fire, as shown in the sketch.

315 g/s Air 285 g/s

Fuel 10 g/s

Data: Room volume

28.8 m3

Ambient temperature

35 C

Ambient density

1.18 kg/m3

Heat of combustion

24 kJ/g

Fuel gas entering temperature

300 C

Exiting gas temperature

275 C

Exiting oxygen mass fraction

0.11

Specific heat at constant pressure

1.3 J/g K (same for all species)

Assume constant pressure throughout and a control volume of the room gas can be considered to enclose uniform properties.

74

CONSERVATION LAWS FOR CONTROL VOLUMES (a) Compute the time rate of change of the room gas density. (b) Compute the rate of heat transfer from the room gases and state whether it is a loss or a gain.

3.14

Heptane spills on a very hot metal plate causing its hot vapor to mix with air. At some point the mixture of heptane vapor and air reaches a condition that can cause autoignition. In the sketch, T is the temperature of the mixture and YF is the mass fraction of heptane vapor. YF T

Hot metal plate

Autoignition is to be investigated by examining local conditions at a small volume in the mixture to see what temperature and fuel mass fraction are needed in the immediate surroundings (T1 and YF;1 ) to just allow ignition. Here we will assume that the small volume of fluid enclosed in a control volume has uniform properties: , density; T, temperature; YF , fuel mass fraction; and cp the specific heat and constant for all species. In addition, the pressure is constant throughout, and heat and fuel mass diffusion occurs between the reacting fluid in the volume and its surroundings at fixed conditions (T1 and YF;1 ). Control volume

ρ , T , YF

T∞,YF,∞

For the control volume, the heat flux at the boundary is given as q_ 00 ¼ hc ðT T1 ). The diffusion mass flux supplying the reaction is given as m_ 00F ¼ hm ðYF;1 YF ), where from heat and mass transfer principles hm ¼ hc =cp . Let V and S be the volume and surface area of the E=ðRTÞ control volume. The reaction rate per unit volume is given as m_ 000 for the fuel F ¼ AYF e in this problem. Perform the following: (a) Write the conservation of mass, fuel species and energy for this control volume. Show your development. (b) Show, for fuel mass diffusion as negligible in the mass and energy conservation equations (only), that the conservation of species (fuel) and energy can be written as dYF S hc ¼ AYF eE=ðRTÞ þ YF;1 YF dt cp V dT S ¼ hc AYF eE=ðRTÞ h c ðT T 1 Þ cp dt V

PROBLEMS

75

(c) Show under steady conditions that YF ¼ YF;1

cp ðT T1 Þ hc

(Hint: use part (b) here.) (d) Consider YF;1 to be at its stoichiometric value for a mixture with air. For heptane, C7H16, going to complete combustion, calculate YF;1 from the stoichiometric equation. (e) Using a graphical, computational analysis for heptane, compute the following (use the result of part (b) and the data below): (i) T1 , the minimum temperature to allow autoignition; (ii) YF , the fuel mass fraction ignited; (iii) identify the points of steady burning and compute the flame temperature for the surroundings maintained at T1 of part (e)(i). Assume a constant volume sphere that reacts having a radius (r) of 1 cm (V ¼ 43r 3 ; S ¼ 4r 2 ) using the following data: Preexponential, A ¼ 9:0 109 g=m3 s Activation energy, E ¼ 160 kJ/mol (R ¼ 8:314 J=mol K) Heat of combustion, hc ¼ 41:2 kJ/g Specific heat, cp ¼ 1:1 J/g K Conductivity, k ¼ 0:08 W/m K Heat transfer coefficient, hc k=r Surrounding fuel mass fraction, YF;1 ¼ 0:069

4 Premixed Flames

4.1 Introduction A premixed flame is a chemical reaction in which the fuel and the oxygen (oxidizer) are mixed before they burn. In a diffusion flame the fuel and oxygen are separated and must come together before they burn. In both cases we must have the fuel and the oxidizer together in order to burn. Thermodynamics can tell what happens when they burn, provided we know the chemical equation. We can place a gaseous fuel and oxygen together, yet nothing may occur. Even the requirements of chemical equilibrium from thermodynamics are likely not to ensure the occurrence of a chemical reaction. To answer the question of whether a reaction will occur, we must understand chemical kinetics, the subject related to the rate of the reaction and its dependent factors. We shall examine several aspects of premixed flames: 1. What determines whether a reaction occurs? 2. What is the speed of the reaction in the fuel–oxidizer mixture? 3. What determines extinction of the reaction? We shall discuss only gas mixtures, but the results can apply to fuel–oxidizer mixtures in liquid and solid form. The fuel in an aerosol state can also apply. At high speeds, compressibility effects will be important. At reaction front speeds greater than the speed of sound, a shock front will precede the reaction front. Such a process is known as a detonation, compared to reaction front speeds below sonic velocity which are called deflagrations. We shall only investigate the starting speed under laminar conditions. Beyond this point, one might regard the subject as moving from fire to explosion. The subject of explosion is excluded here. However, the means to prevent a potential explosion hazard are not excluded.

Fundamentals of Fire Phenomena James G. Quintiere # 2006 John Wiley & Sons, Ltd ISBN: 0-470-09113-4

78

PREMIXED FLAMES

Figure 4.1

Premixed and diffusion flames

Although fire mainly involves the study and consequences of diffusion flames, premixed flames are important precursors. In order to initiate a diffusion flame, we must first have a premixed flame. In regions where a diffusion flame is near a cold wall, we are likely to have an intermediary premixed flame. Even in a turbulent diffusion flame, some state of a premixed flame must exist (see Figure 4.1).

4.2 Reaction Rate There are many molecular and atomic mechanisms for the promotion of a chemical reaction. A catalyst is a substance that does not take part in the reaction, but promotes it. A radical is an unstable species seeking out electrons, as molecules are colliding and bonding in a complex path to the end of the reaction. The temperature is directly related to the kinetic energy of the molecules and their collision rate. This microscopic description is very approximately described by an empirical global kinetic rate equation. The basis for it is attributed to Suante Arrhenius [1], who showed that the thermal behavior for the rate is proportional to eE=ðRTÞ , where R is the universal gas constant and E is called the activation energy having units of kJ/mole. E=ðRTÞ is dimensionless. The significance of E is that the lower its value, the easier it is for a chemical reaction to be thermally initiated. Despite the apparent restriction by R to the perfect gas law, the relationship is also used to describe chemical rates in solids and liquids. Intuitively, the rate for a chemical reaction must depend on the concentrations of fuel and oxidizer in the mixture. An overall reaction rate expression, which is often used to match the effect of the complex reaction steps needed, is given by n m E=ðRTÞ m_ 000 F;r ¼ A0 YF YO2 e

ð4:1aÞ

79

REACTION RATE Table 4.1 Overall reaction rates for fuels burning in air (from Westbrook and Dryer [2])a Fuel Methane Ethane Propane n-Hexane n-Heptane Octane Methanol Ethanol a

Formula CH4 C2 H6 C3 H8 C6 H14 C7 H16 C8 H18 CH3 OH C2 H5 OH

B

E(kcal/mole)

n

m

30.0 30.0 30.0 30.0 30.0 30.0 30.0 30.0

0.3 0.1 0.1 0.25 0.25 0.25 0.25 0.15

1.3 1.65 1.65 1.5 1.5 1.5 1.5 1.6

5

8:3 10 1:1 1012 8:6 1011 5:7 1011 5:1 1011 4:6 1011 3:2 1012 1:5 1012

Here the rate constant is given by d½F ¼ k½F n ½O2 m dt

where [ ] implies molar concentration of the species in mole/volume, k B eE=ðRTÞ This compares to Equation (4.1a) with A0 ¼

Bnþm1 MFn1 MOm2

or E=ðRTÞ m_ 000 F;r ¼ A e

ð4:1bÞ

This is commonly called the Arrhenius equation. Table 4.1 gives typical values for fuels in terms of the specific rate constant, k. In Equation (4.1), m_ 000 F;r is taken as positive for the mass rate of fuel consumed per unit volume. Henceforth, in the text we will adopt this new sign convention to avoid the minus sign we were carrying in Chapter 3. The quantity A is called the pre-exponential factor and must have appropriate units to give the correct units to m_ 000 F;r . The exponents n and m as well as A must be arrived at by experimental means. The sum ðn þ mÞ is called the order of the reaction. Often a zeroth-order reaction is considered, and it will suffice for our tutorial purposes. Illustrative values for a zeroth-order reaction are given as follows: A ¼ 1013 g=m3 s and E ¼ 160 kJ=mole (This value for A is somewhat arbitrary since it could range from about 105 to 1015 .) Table 4.2 accordingly gives a range of values for m_ 000F;r. The tabulation shows a strong sensitivity to temperature. It is clear that between 298 and 1200 K there is a significant change. In energy units, for a typical heat of combustion of a gaseous fuel of 45 kJ/g, we have about 50 106 or 50 kJ=cm3 s at 1200 K. One calorie or 4.184 J can raise 1 cm3 of

80

PREMIXED FLAMES Table 4.2

Typical fuel production for a zeroth-order arrhenius rate

TðKÞ

E=ðRTÞ

exp½E=ðRTÞ

m_ 000F;r ðg=m3 sÞ

Energy release ðW=cm3 Þ

298 600 1200 1600 2000 2500

64.6 32.1 16.0 12.0 9.6 7.7

9:0 1030 1:18 1014 1:08 107 5:98 106 6:62 105 4:54 104

9:0 1017 1:2 101 1:1 106 6:0 107 6:6 108 4:5 109

0:40 1017 0:53 102 0:53 105 0:27 107 0:30 108 0:20 109

water 1 K. Here we have 50 000 J released within 1 cm3 for each second. At 298 K this is imperceptable (0:4 1017 J=cm3 s) and at 600 K it is still not perceptible (0:54 102 J=cm3 s). Combustion is sometimes described as a chemical reaction giving off significant energy in the form of heat and light. It is easy to see that for this Arrhenius reaction, representative of gaseous fuels, the occurrence of combustion by this definition might be defined for some critical temperature between 600 and 1200 K.

4.3 Autoignition Raising a mixture of fuel and oxidizer to a given temperature might result in a combustion reaction according to the Arrhenius rate equation, Equation (4.1). This will depend on the ability to sustain a ‘critical’ temperature and on the concentration of fuel and oxidizer. As the reaction proceeds, we use up both fuel and oxidizer, so the rate will slow down according to Arrhenius. Consequently, at some point, combustion will cease. Let us ignore the effect of concentration, i.e. we will take a zeroth-order reaction, and examine the concept of a ‘critical’ temperature for combustion. We follow an approach due to Semenov [3]. A gaseous mixture of fuel and oxidizer is placed in a closed vessel of fixed volume. Regard the vessel as a perfect conductor of heat with negligible thickness. The properties of the gaseous mixture are uniform in the vessel. Of course, as the vessel becomes large in size, motion of the fluid could lead to nonuniformities of the properties. We will ignore these effects. Assuming constant and equal specific heats, we adopt Equation (3.45) for a control volume enclosing the constant volume, fixed mass vessel (Figure 4.2):

mcv

dT ¼ Q_ þ ðm_ 000 F;r ÞðVÞðhc Þ dt

ð4:2Þ

The heat added term is given in terms of a convective heat transfer coefficient, h, and surface area, S, Q_ ¼ hSðT1 TÞ

ð4:3Þ

AUTOIGNITION

Figure 4.2

81

Autoignition in a closed vessel

Actually, we expect T > T1 due to the release of chemical energy, so this is a heat loss. We write this loss as Q_ L ¼ hSðT T1 Þ

ð4:4Þ

The chemical energy release rate is written as Q_ R ¼ m_ 000 F;r V hc

ð4:5Þ

Then we can write mcv

dT ¼ Q_ R Q_ L dt

ð4:6Þ

Figure 4.3 tries to show the behavior of these terms in which Q_ R is only approximately represented for the Arrhenius dependence in temperature. For a given fuel and its associated kinetics, Q_ R is a unique function of temperature. However, the heat loss term depends on the surface area of the vessel. In Figure 4.3, we see the curves for increasing

Figure 4.3

Competition of heat loss and chemical energy release

82

PREMIXED FLAMES

size vessels ðS1 ; S2 ; S3 Þ for a constant h. Where we have no intersection of Q_ L and Q_ R , Q_ R > Q_ L and by Equation (4.6) dT=dt > 0. This means that the temperature will continue to rise at an increasing rate by the difference between the curves. This is usually called thermal runaway or a thermal explosion. On the other hand, when the curves intersect, we must have dT=dt ¼ 0 by Equation (4.6) or a steady state solution. Two intersections are possible for the zeroth-order reaction. The one labeled s is stable, while that labeled u is unstable. This can be seen by imposing small changes in temperature about u or s. For example, a small positive temperature perturbation at s gives Q_ L3 > Q_ R so by Equation (4.6) dT=dt < 0, and the imposed temperature will be forced back to that of s. Likewise small changes of temperature about point u will either cause a runaway ðþTÞ or a drop to s ðTÞ. The critical condition is given by point c with Tc being the critical temperature. It should be noted that the critical heat loss rate, Q_ L2 , depends not just on the vessel size but also on the surrounding temperature, T1 . Hence, the slope and the T intersection of Q_ L in Equation (4.4), as tangent to Q_ R , will give a different Tc . We shall see that the Arrhenius character of the reaction rate will lead to Tc T1 . This is called the autoignition temperature. The mathematical analysis is due to Semenov [3]. An approximate mathematical analysis is considered to estimate the functional dependence of the gas properties and vessel size on the autoignition temperature. We anticipate that Tc will be close to T1 , so we write E T1 eE=ðRTÞ ¼ exp RT1 T T1 þ T 1 " 1 # E T T1 þ1 ¼ exp RT1 T1

ð4:7Þ

Since ðT T1 Þ=T1 is expected to be small, we approximate 1 T T1 T T1 þ1 1 T1 T1

ð4:8Þ

where these are the first two terms of an infinite power series. Then, defining

T T1 E RT1 T1

eE=ðRTÞ e eE=ðRT1 Þ

ð4:9Þ

ð4:10Þ

from Equations (4.7) to (4.9). Since is dimensionless, it is useful to make Equation (4.2) dimensionless. By examining the dimensions of Vcv

dT

hSðT T1 Þ dt

AUTOIGNITION

83

a reference time can be defined as tc ¼

Vcv hS

ð4:11Þ

Physically, tc is proportional to the time required for the gases in the vessel to lose their stored energy to the surroundings by convection. Let us define a dimensionless time as ¼

t tc

ð4:12Þ

Now Equation (4.2) can be written with the approximation for eE=ðRTÞ as d ¼ e d

ð4:13Þ

where ¼

hc VAeE=ðRT1 Þ hST1

E RT1

(4.14)

is a dimensionless quantity. The equation can be solved with ¼ 0 at ¼ 0 to give the temperature rise for a given . The increase in from its initial value is directly related to , a Damkohlar number. We are only interested here in whether combustion will occur. From Figure 4.3, we need to examine point c. In terms of Equation (4.13), we have the dimensionless forms of Q_ R and Q_ L on the right-hand side. We plot two quantities, and e , against in Figure 4.4. The critical condition c is denoted by the tangency of the two quantities. Mathematically, this requires that each quantity and their slopes must be equal at c. Hence, c ec ¼ c

ð4:15Þ

c e c ¼ 1

ð4:16Þ

and

Figure 4.4

Conditions for autoignition

84

PREMIXED FLAMES

Dividing Equation (4.15) by Equation (4.16) gives c ¼ 1

ð4:17Þ

c ¼ e1

ð4:18Þ

and then it follows that

For a given vessel size, Equations (4.14) and (4.18) give the minimum value of T1 to cause autoignition. This value, T1;c , is the minimum temperature needed, and is called the autoignition temperature. The dependence on the chemical parameters, A, E and hc , as well as size are clearly contained in the parameter, . Also, from Equations (4.9) and (4.17), Tc T1 RT1 ¼ T1 E

ð4:19Þ

which according to Table 4.1 suggests a value for Tc of roughly 1=30 above a T1 of 600 K, i.e. Tc T1 ð8:314 J=mole KÞð600 KÞ 1 ¼ ¼ 3 ð160 10 J=moleÞ 32:1 T1

ð4:20Þ

or Tc ¼ 618:7 K. As we can see, the value Tc is very close to the value T1;c needed to initiate the combustion. Practical devices, representative of the system in Figure 4.2, can regulate T1 until ignition is recognized, e.g. ASTM E 659 [4]. Table 4.3 gives some typical measured autoignition temperatures (AIT). There is a tendency for the AIT to drop as the pressure is increased.

Table 4.3 Autoignition temperatures (at 25 C and 1 atm, usually at stoichiometric conditions in air) (from Zabetakis [5]) Fuel

AIT ( C)

Methane, CH4 Ethane, C2 H6 Propane, C3 H8 n-Hexane, C6 H14 n-Heptane, C7 H16 Methanol, CH3 OH Ethanol, C2 H5 OH Kerosene Gasoline JP-4 Hydrogen, H2

540 515 450 225 215 385 365 210

450 240 400

PILOTED IGNITION

85

4.4 Piloted Ignition Often combustion is initiated in a mixture of fuel and oxidizer by a localized source of energy. This source might be an electric arc (or spark) (moving charged particles in a fluid or a plasma) or a small flame itself. Because the spark or small flame would locally raise the temperature of the mixture (as T1 did in the autoignition case), this case is defined as piloted ignition. The bulk of the mixture remains at T1 , well below the AIT. To illustrate the process of piloted ignition, we consider a cylindrical spark of radius, r, and length, ‘, discharged in a stoichiometric mixture of fuel and air. A control volume enclosing the spark is shown in Figure 4.5 where it can expand on heating to match the ambient pressure, p1 . The governing energy equation for this isobaric system is taken from Equation (3.45), where p ¼ p1 , m_ j ¼ 0 and Vcp

dT _ 000 V hSðT T1 Þ ¼ m_ 000 F hc V þ q dt

ð4:21Þ

where q_ 000 is the spark rate of energy supplied per unit volume and h is an effective conductance, k=r, with k as the mixture thermal conductivity. This is an approximate analysis as in Equations (4.2) to (4.5), except that we have added the spark energy and we approximate a conduction rather than a convective heat loss. By introducing the same dimensionless variables, it can be shown that 000 d E q_ V ¼ e þ d hST1 RT

ð4:22Þ

where here tc ¼ Vcp =ðhSÞ ¼ r 2 =ð2Þ with , the thermal diffusivity, k=cp , and V ¼ r 2 ‘ and S ¼ 2 r‘ as approximations. By examining the definition of , Equation (4.14), it can be recognized as the ratio of two times ¼

tc tR

ð4:23Þ

where tc is the time needed for the conduction loss, or thermal diffusion time, and tR is the chemical reaction time: tR

cp T1 ½E=ðRT1 Þ hc A eE=ðRT1 Þ

Figure 4.5

Spark ignition

(4.24)

86

PREMIXED FLAMES Table 4.4 Minimum spark energy to ignite stoichiometric mixtures in air (at 25 C, 1 atm) (from various sources) Fuel

Energy (mJ)

Methane n-Hexane Hydrogen

300 290 17

The critical solution follows as in Equations (4.15) to (4.18): c ¼ 1 þ

q_ 000 V hST1

E RT

¼1þ

q_ 000 r 2 2kT1

E RT

ð4:25aÞ

and c ¼ ec ¼ e1 eðq_

000 2

r =ð2kT1 ÞÞðE=ðRTÞÞ

ð4:25bÞ

Hence the minimum value of to cause ignition is smaller for a system under the heating of a ‘pilot’ compared to autoignition, e.g. Equation (4.18). Analogous to Figure 4.4, the point of tangency, c, corresponds R to a minimum spark energy. This has usually been measured in terms of energy ð q_ 000 V dtÞ for the spark duration. Table 4.4 gives some typical spark energies. While the minimum electric spark energy is generally found to range from 10 to 1000 mJ, the power density of a spark in these experiments is greater than 1 MW/cm3. Figure 4.6 shows the minimum spark energy needed to ignite methane over a range of flammable mixtures. Within the

Figure 4.6 Ignitability curve and limits of flammability for methane-air mixtures at atmospheric pressure and 26 C (taken from Zabetakis [5])

PILOTED IGNITION

87

U-shaped curve, we have mixtures that can be ignited for a sufficiently high spark energy. From Equation (4.25) and the dependence of the kinetics on both temperatures and reactant concentrations, it is possible to see why the experimental curve may have this shape. The lowest spark energy occurs near the stoichiometric mixture of XCH4 ¼ 9:5 %. In principle, it should be possible to use Equation (4.25) and data from Table 4.1 to compute these ignitability limits, but the complexities of temperature gradients and induced flows due to buoyancy tend to make such analysis only qualitative. From the theory described, it is possible to illustrate the process as a quasi-steady state ðdT=dt ¼ 0Þ. From Equation (4.21) the energy release term represented as Q_ R ¼ m_ 000 F hc V

ð4:26Þ

Q_ L ¼ hSðT T1 Þ q_ 000 V

ð4:27Þ

and the net loss term as

can be sketched as a function of the reacting mixture temperature. In this portrayal, we are ignoring spatial gradients except for their effect in the heat loss part of Q_ L . As shown in Figure 4.7, Q_ R will first sharply increase due to the Arrhenius temperature dependence, eE=ðRTÞ , but later it will diminish due to the reduction of the reactants. The net loss term can be approximated as linear in T, with T1 representative of an ambient starting

Figure 4.7

Qualitative description of the minimum energy needed for ignition

88

PREMIXED FLAMES

temperature (e.g. 26 C and 1 atm as in Figure 4.6). The net loss curve labeled 1 has a stable intersection corresponding to Tf , a flame temperature. For this protrayal, the reactants would have to be continuously supplied at a concentration corresponding to this intersection. If they were not replenished, the fire would go out as Q_ R goes to zero. Curve 1 corresponds to an ignition (spark) energy greater than the minimum value. Curve 1 gives the critical ignition result where two solution temperatures, Tc and Tf;c are possible. In the dynamic problem dT=dt is always greater than zero because Q_ R > Q_ L at the Tc neighborhood. Hence, the steady physically plausible solution will be the fire condition at Tf;c . Any lower value of ðq_ 000 VÞ will result in an intersection corresponding to T0 , a very small and imperceptible difference from T1 (see curve 0). This intersection is stable and not a state of combustion. Here, it should be emphasized that the coordinate scales in Figure 4.7 are nonlinear (e.g. logarithmic), and quantitative results can only be seen by making computations. In dimensionless terms, there is a critical value for (Damkohler number) that makes ignition possible. From Equation (4.23), this qualitatively means that the reaction time must be smaller than the time needed for the diffusion of heat. The pulse of the spark energy must at least be longer than the reaction time. Also, the time for autoignition at a given temperature T1 is directly related to the reaction time according to Semenov (as reported in Reference [5]) by E logðtauto Þ ¼ þb ð4:28Þ RT1 where =R ¼ 0:22 cal/mole K and b is an empirical constant for the specific mixture. Theoretically, b would have some small dependency on T1 , i.e. logðT1 Þ, as would follow from Equation (4.24). Thus, tauto tR and tspark tR for piloted ignition.

4.5 Flame Speed, Su Flammability is defined for a mixture of fuel and oxidizer when a sustained propagation occurs after ignition. This result for a given mixture depends on temperature, pressure, heat losses and flow effects, namely due to gravity. The speed of this propagation also depends on whether the flow remains laminar or becomes turbulent, and whether it exceeds the speed of sound. For speeds below the speed of sound for the medium, the process is called a deflagration. The pressure rise across the burning region is relatively low for a deflagration. However, for a detonation – speeds greater than the speed of sound – the pressure rise is considerable. This is the result of a shock wave that must develop before the combustion region in order to accommodate the conservation laws. We will not discuss detonations in any detail. However, we will discuss the initiation of a deflagration process in a gas mixture that can ideally be regarded to occur without heat losses. This ideal flame speed is called the burning velocity and is designated by Su. Specifically, Su is defined as the normal speed of an adiabatic plane (laminar) combustion region measured with respect to the unburned gas mixture. Su , in this idealization, can be treated as a property of the fuel– oxidizer mixture, but it can only be approximately measured. Typical measured values of Su are shown in Figure 4.8 which peak for a slightly fuel-rich mixture. Typical maximum values of Su (25 C, 1 atm) in air

FLAME SPEED, Su

89

Figure 4.8 Typical burning velocities (taken from Zabetakis [5])

are slightly less than 1 m/s with acetylene at 1.55 m/s and hydrogen at 2.9 m/s (which actually peaks at an equivalence ratio of 1.8). In contrast detonation speeds can range from 1.5 to 2.8 km/s. Turbulence and pressure effects can accelerate a flame to a detonation.

4.5.1

Measurement techniques

Figures 4.9(a) and (b) show two measurement strategies for measuring Su . One is a Bunsen burner and the other is a burner described by Botha and Spalding [6]. In the

Figure 4.9

Burners to measure Su

90

PREMIXED FLAMES

Bunsen burner the conical flame angle ð2Þ and the mean unburned mixture velocity ðvu Þ in the tube can approximately give Su : Su ¼ vu sin

ð4:29Þ

In the Spalding burner, a flat plane flame stands off the surface of a cooled porous matrix. Data are compiled to give Su as measured by the speed in the burner supply, to maintain a stable flame, for a given measured cooling rate. By plotting these data so as to extrapolate to a zero cooling condition yields Su under nearly adiabatic conditions. In both of these experimental arrangements, for a given mixture, there is a unique duct velocity ðvu Þ that matches the burning velocity. In the Spalding burner, this is the adiabatic burning velocity (or the true Su ). If vu > Su the condition is not stable and the flame will blow off or move away from the exit of the duct until a reduced upstream velocity matches Su . If vu < Su , the flame will propagate into the duct at a speed where the flame velocity is Su vu . This phenomenon of upstream propagation is known as flashback. It is not a desirable effect since the flame may propagate to a larger chamber of flammable gases and the larger energy release could result in a destructive pressure rise. An appropriate safety design must avoid such a result.

4.5.2

Approximate theory

The ability to predict Su is limited by the same factors used to predict the autoignition or flammability limits. However, an approximate analysis first considered by Mallard and Le Chatelier in 1883 [7] can be useful for quantitative estimates. Consider a planar combustion region in an adiabatic duct that is fixed in space and is steadily supplied with a mixture of fuel (F), oxidizer (O) and diluent (D) at the velocity Su . This condition defines Su and is depicted in Figure 4.10. The process is divided into two stages: I. A preheat region in which the heat transfer from the flame brings the unburned mixture to its critical temperature for ignition, Tig . This is much like what occurred in describing auto and piloted ignition, except that the the heat is supplied from the flame itself. II. The second stage is the region where a significant chemical energy is released. This is perceived as a flame. The boundary between these two regions is not sharp, but it can be recognized. We apply the conservation laws to two control volumes enclosing these regions. Since there is no change in area, conservation of mass (Equation (3.15)) gives, for the unburned mixture (u) and burned product (b), u Su ¼ b vb ¼ m_ 00 or the mass flux, where m_ 00 is constant.

ð4:30Þ

FLAME SPEED, Su

Figure 4.10

91

Burning velocity theory

The conservation of energy applied to CVI requires a knowledge of the ‘ignition’ 00 temperature, Tig and the heat transferred to the preheat region, q_ . We assume that radiation effects are negligible and approximate this heat flux as Tb Tig dT k q_ ¼ k dx R 00

ð4:31Þ

We are only seeking an approximate theory since a more precise simple analytical result is not possible. We only seek an order of magnitude estimation and insight into the important variables. With this approximation the energy equation for CVI is u Su cp ðTig Tu Þ k

Tb Tig R

ð4:32Þ

Regarding properties as constants, evaluated at the unburned mixture condition, Equations (4.30) and (4.32) combine to give Tb Tig Su Tig Tu R

(4.33)

where the thermal diffusivity, , is k=cp . As an order of magnitude estimate, we can regard Tig as a typical autoignition temperature 525 C, Tu 25 C and Tb as an

92

PREMIXED FLAMES

adiabatic flame temperature 2000 C. Therefore, Tb Tig 1500 ¼3 500 Tig Tu However, Su is typically 0.5 m/s and is typically 2 105 m2 =s. Hence, the thickness of the flame is estimated as R ð3Þ

ð2 105 m2 =sÞ ¼ 12 105 m 104 m ð0:5 m=sÞ

or the flame is roughly of the order of 0.1 mm thick. This would be consistent with the laminar flame of a Bunsen burner or oxyacetylene torch. To gain further insight into the effect of kinetics on the flame speed, we write the energy equation for the control volume CVII as _ 00 u Su cp ðTb Tig Þ ¼ m_ 000 F hc R q

ð4:34Þ

We can make the same approximation for q_ 00, as in Equation (4.31), and substitute for R using Equation (4.33). It can be shown that Su ¼

½ðTb Tig Þ=ðTig Tu Þ m_ 000 F hc u cp ðTb Tu Þ

1=2 ð4:35Þ

As before, since Tig is not precisely known, we approximate Tb Tig 1500 ¼3 500 Tig Tu and substitute to estimate the flame speed as Su

3hc m_ 000 F u cp ðTb Tu Þ

1=2 (4.36)

The constant 3 is an order of magnitude estimate which depends on the approximate analysis for the model of Figure (4.10).* In addition, the burning rate m_ 000 F and the properties and cp should be evaluated at some appropriate mean temperature. For example, in Equation (4.34), the more correct expression for m_ 000 F R is Z 0

R

m_ 000 F dx ¼

Z

Tb

Tig

m_ 000 F dT dT=dx

* An asymptotic analysis for a large activation energy or small reaction zone obtains 2 instead of 3 for the constant [8].

FLAME SPEED, Su

93

or R Tb m_ 000 F R

4.5.3

Tig

m_ 000 F dT

Tb Tig

R

Fuel lean results

For our estimations and the adiabatic control volume in Figure 4.10, Tb should be the adiabatic flame temperature. Consider a fuel-lean case in which no excess fuel leaves the control volume. All the fuel is burned. Then by the conservation of species, u Su YF;u ¼ m_ 000 F R

ð4:37Þ

In addition, conservation of energy for the entire control volume ðI þ IIÞ gives u Su cp ðTb Tu Þ ¼ m_ 000 F R hc

ð4:38Þ

Combining these equations gives cp ðTb Tu Þ ¼ YF;u hc

ð4:39Þ

for the fuel-lean adiabatic case. Then Equation (4.36) can be written as 1=2 3m_ 000 F Su ¼ u YF;u

(4.40)

where this applies to fuel-lean conditions and Tb is the adiabatic flame temperature given by Equation (4.39). These results show that the flame speed depends on kinetics and thermal diffusion. Any turbulent effects will enhance the thermal diffusion and the surface area so as to increase the flame speed. Any heat losses will decrease the temperature and therefore the burning rate and flame speed as well. Indeed, if the heat losses are sufficient, a solution to Equation (4.38) may not be possible. Let us investigate this possibility.

4.5.4

Heat loss effects and extinction

Consider the entire control volume in Figure 4.10, but now heat is lost to the surroundings at the temperature of the unburned mixture, Tu . Such an analysis was described by Meyer [9]. The conservation of energy for the entire control volume is _ 00 u Su cp ðTb Tu Þ ¼ m_ 000 F R hc q

ð4:41Þ

Consider the fuel-lean case, in which Equation (4.37) applies. Substituting gives m_ 000 F R

cp ðTb Tu Þ hc ¼ q_ 00 YF;u

ð4:42Þ

94

PREMIXED FLAMES

Figure 4.11

Solution for steady flame speed

Note that when q_ 00 ¼ 0, Tb is the adiabatic flame temperature. We can regard this equation as a balance between the net energy released and energy lost: _QR 00 ¼ m_ 000 R hc cp ðTb Tu Þ F YF;u

ð4:43aÞ

and kðTb Tu Þ Q_ L 00 R

or

ðTb Tu Þ

ð4:43bÞ

which approximates the heat loss from a linear decrease in the product temperature.* The flame thickness can be approximated from Equation (4.33) as R

3 Su

ð4:44Þ

and Su taken from Equation (4.36). Then the steady flame temperature can be formed by equating Equations (4.43a) and (4.43b). Figure 4.11 contains a qualitative sketch of the Q_ 00R and Q_ 00L curves as a function of Tb . In the fuel-lean range, YF;u is greatest at the stoichiometric mixture. For such a case, an intersection of the curves gives a steady propagation solution, depicted as 1. The *

The heat losss from the combustion products in Figure 4.10 is due to conduction and radiation.

QUENCHING DIAMETER

95

temperature is below the adiabatic flame temperature for the stoichiometric mixture. As YF;u is decreased, a point of tangency will occur for some value. This is the lowest fuel concentration for which a steady propagation is possible. Any further decrease in YF;u will not allow propagation even if ignition is achieved. The critical tangency condition value for this fuel concentration is called the lower flammable limit (LFL). This condition is depicted as 2 in Figure 4.11, and the intersection with the zero energy horizontal axis is the adiabatic flame temperature at the lower limit. This analysis is analogous to the autoignition problem defined by Equation (4.5), in which here m00 cp

dT ¼ Q_ 00R Q_ 00L dt

ð4:45Þ

and at any time Q_ 00R < Q_ 00L and the temperature decreases. At the lower limit, any perturbation to increase heat losses will lead to extinction. Hence, we see that there is a lower limit concentration, on the lean side, below which steady propagation is not possible. As the fuel concentration is increased above stoichiometric on the rich side, the incompleteness of combustion will reduce hc . Hence this will decrease T and the burning rate, causing a similar effect and leading to extinction. This critical fuel-rich concentration is called the upper flammable limit (UFL).

4.6 Quenching Diameter In the previous analysis, the heat loss was taken to be from the flame to the surrounding gas. If a solid is present, it will introduce some heat loss. This will be examined in terms of flame propagation in a duct. For a large-diameter duct, a moving flame may only be affected at its edges where heat loss can occur to the duct wall. This affected region is called the quenching distance. For a small-diameter duct, the duct diameter that will not allow a flame to propagate is called the quenching diameter, DQ . The process of steady flame propagation into a premixed system is depicted in Figure 4.12 for a moving control volume bounding the combustion region R . The heat loss in this case is only considered to the duct wall. With h as the convection heat transfer coefficient, the loss rate can be written as Q_ L ¼ h D R ðTb Tw Þ

Figure 4.12

Quenching distance analysis

ð4:46Þ

96

PREMIXED FLAMES

where Tb is approximated as uniform over R and Tw is the wall temperature. For fully developed laminar flow in a duct of diameter D, the Nusselt number is a constant of hD ¼ 3:66 k

ð4:47Þ

according to the heat transfer literature [10]. (Note that in fully developed flow the velocity profile is parabolic and zero at the wall, so viscous effects will also have a bearing on the flame speed. We are ignoring this profile effect in our one-dimensional model.) The energy balance for this problem can be expressed by Equation (4.45) with Q_ 00R given in Equation (4.43a) as before and Q_ 00L given from Equations (4.47) and (4.48) as D R _Q00 ¼ 3:66 k ðTb Tw Þ L D D2=4 R ðTb Tw Þ ¼ 14:64 k ð4:48Þ D2 Again R follows from Equation (4.44) and Su from Equation (4.36). The equilibrium solutions for steady propagation are sketched in Figure 4.13. Only the loss curves depend on D. For D > DQ, two intersections are possible at s and u. The former, s, is a stable solution since any perturbation in temperature, Tb , will cause the state to revert to s. For example, if Tb > Tb;s , Q_ 00L > Q_ 00R and dT=dt < 0 from Equation (4.45). Then Tb will decrease until Tb;s is reached. The intersection at u can be shown to be unstable, and therefore physically unrealistic. The point Q corresponds to the smallest diameter that will still give a stable solution. For D < DQ, dT=dt < 0 always, and any ignition will not

Figure 4.13

Quenching distance solution

97

QUENCHING DIAMETER Table 4.5 Flame speed and quenching diameter Fuel Hydrogen Acetylene Methane

Su , Maximum (in air) (m/s)

DQ, Stoichiometric (in air) (mm)

2.9 1.5 0.52

0.64 2.3 2.5

lead to an appropriate temperature for propagation. Any increase in wall temperature, Tw > Tu , will have a tendency to enable propagation. By equating Q_ 00R and Q_ 00L , with the following approximations a relationship between Su and DQ is suggested. From the sketch in Figure 4.13, the positive term in Q_ 00R is more dominant near the solution, Q, so let Tb Tf;ad . Therefore Q_ 00R m_ 000 F R hc and for Tw ¼ Tu, ðTf;ad Tu Þ R Q_ 00L 14:64 k D2Q Then DQ

14:64 kðTf;ad Tu Þ m_ 000 F hc

1=2 ð4:49Þ

From Equation (4.36), DQ

6:6 Su

ð4:50Þ

This seems to qualitatively follow data in the literature, as illustrated in Table 4.5. The quenching distance is of practical importance in preventing flashback by means of a flame arrestor. This safety device is simply a porous matrix whose pores are below the quenching distance in size. However, care must be taken to maintain the flame arrestor cool, since by Equation (4.49) an increase in its temperature will reduce the quenching distance. The quenching distance is also related to the flame stand-off distance depicted in Figure 4.14. This is the closest distance that a premixed flame can come to a surface.

Figure 4.14

Stand-off distance

98

PREMIXED FLAMES

Even in a diffusion flame the region near a surface becomes a premixed flame since fuel and oxygen can come together in this ‘quenched’ region.

4.7 Flammability Limits In Figure 4.6 there are two apparent vertical asymptotics where no ignitable mixture is possible regardless of the size of the ignition energy source. These limits given in concentration of fuel are defined as the lower ðXL Þ and upper ðXU Þ flammability limits. They are typically given in terms of molar (or volume) concentrations and depend for a given mixture (e.g. fuel in air or fuel in oxygen and an additive species) on temperature and pressure. This demarcation is also dependent on the nature of the test apparatus and conditions of the test in practice. Generally, data are reported from a standard test developed by the US Bureau of Mines [11], which consists of a spark igniter at the base of an open 2 inch diameter glass tube. Flammability is defined for a uniformly distributed mixture if a sustained flame propagates vertically up the tube. Outside the flammable regions, a spark may succeed in producing a flame, but by the definition of flammability, it will not continue to propagate in the mixture. Since flammability, as ignitability, can depend on flow and heat transfer effects, it cannot be considered to be a unique property of the mixture. It must be regarded as somewhat dependent on test conditions. Nevertheless, reported values for upper and lower flammability limits have a useful value in fire safety design considerations and can be used with some generality. A useful diagram to show the effect of fuel phase and mixture temperature on flammability is shown in Figure 4.15. The region of autoignition is characterized as a combustion reaction throughout the mixture at those temperatures and concentrations indicated. Outside that temperature domain only piloted ignition is possible, and flammability is manifested by a propagating combustion reaction through the mixture. At temperatures low enough, the gaseous fuel in the mixture could exist as a liquid. The saturation states describe the gaseous fuel concentration in equilibrium with its liquid. In other words, this is the vapor concentration at the surface of the liquid fuel, say in air, at the given temperature. Recall that this saturation (or mixture partial pressure) is only a function of temperature for a pure substance, such as single-component liquid fuels. The lowest flammable temperature for a liquid fuel in air is known as the flashpoint (FP).

Figure 4.15

Effect of temperature on flammability for a given pressure (taken from Zabetakis [5])

FLAMMABILITY LIMITS

Figure 4.16

99

Flammability diagram for three-component systems at 26 C and 1 atm (taken from Zabetakis [5])

However, if the fuel is in the form of droplets in air, it can be flammable at temperatures below the flashpoint. Droplet size will also play a role in this aerosol region. Zabetakis [5] describes two types of flammability diagrams useful to analyze three component mixtures consisting of (1) fuel, (2) oxidizer and (3) inert (diluent) or fire retardant. For example, an inert species could be nitrogen or carbon dioxide which can serve to dilute the fuel upon its addition to the mixture. A fire retardant third body is, for example, carbon tetrafluoride ðCF4 Þ. A fire retardant dilutes as well as inhibits the combustion reaction by playing a chemical role. Either a rectangular or triangular format can be used to represent the flammability of such a three-component system. Figures 4.16(a) and (b) illustrate CH4 O2 N2 . The characteristics of each diagram are: 1. Each point must add to 100 % in molar concentration: XCH4 þ XO2 þ XN2 ¼ 100 %. 2. The concentration of each species at a state on the diagram is read parallel to its zero concentration locus. 3. A species is added by moving towards its vertex at 100 % concentration. 4. The air-line always has O2 and N2 concentrations equal to that of air.

Example 4.1 We take an example from The SFPE Handbook described by Beyler [12]. A methane leak fills a 200 m3 room until the methane concentration is 30 % by volume. Calculate how much nitrogen must be added to the room before air can be allowed in the space. Solution Assume the process takes place at constant temperature and pressure and Figure 4.16 applies. The constant pressure assumption requires that as the methane leaks into the room, the pressure is not increased, as would occur in a constant volume process,

100

PREMIXED FLAMES

i.e. p ¼ mRT=V. Hence, as air is added, the volume of the gases in the room must increase; therefore, these gases leak out of the room (B). We further assume that the mixture in the room is well mixed. In general, this will not be true since methane is lighter than air and will then layer at the ceiling. Under this unmixed condition, different states of flammability will exist in the room. The mixed state is given by B in Figure 4.16(b): B:

XCH4 ¼ 30 %;

XO2 ¼ 0:21ð70Þ %;

XN2 ¼ 0:79ð70Þ %

We add N2 uniformly to the mixture, moving along the B–C line towards 100 % N2 . The least amount of N2 before air can safely be added is found by locating state C just when the addition of air no longer intersects the flammable mixture region. This is shown by the process B ! C and C ! A in Figure 4.16(b): C:

XCH4 ¼ 13 %;

XO2 ¼ 5 %;

XN2 ¼ 82 %

Hence state C is found, but conservation of mass for species N2 must be used to find the quantity of N2 added. Consider the control volume defined by the dashed lines in C. Employ Equation (3.23): m

dYN2 þ m_ N2 ðYN2 1Þ ¼ 0 dt

where YN2 is the mass fraction of the N2 in the mixture of the room, m is the mass of the room and m_ N2 is the mass flow rate of N2 added. There is no chemical reaction involving N2 so the right-hand side of Equation (3.23) is zero. Also, diffusion of N2 has been neglected in the inlet and exit streams because it is reasonable to assume negligible gradients of nitrogen concentration in the flow directions. The mass of the mixture may change during the process B ! C. Expressing it in terms of density ðÞ and volume of the room ðV ¼ 200 m3 Þ, m ¼ V By the perfect gas law for the mixture: ¼

p M RT

or depends only on the molecular weight of the mixture, M¼

3 X

Xi M i

i¼1

For state B: MB ¼ ð0:30Þð16Þ þ ð0:147Þð32Þ þ ð0:543Þð28Þ ¼ 24:7 and for state C: MC ¼ ð0:13Þð16Þ þ ð0:05Þð32Þ þ ð0:82Þð28Þ ¼ 26:6

FLAMMABILITY LIMITS

101

Hence, it is reasonable to approximate the density as constant. With this approximation we can integrate the differential equation. Let y ¼ YN2 1 and YN2 ¼ XN2 ðMN2 =MÞ. Then dy m_ N ¼ 2y dt V where (B) t ¼ 0, YN2 ¼ ð0:543Þð28=24:7Þ ¼ 0:616 and (C) tfinal , YN2 ¼ ð0:82Þ ð28=26:6Þ ¼ 0:863. Thus Rt 0final m_ N2 dt 1 0:863 ln ¼ V 1 0:616 If we represent m_ N2 as a volumetric flow rate of N2 , then the volume of nitrogen added (at p ¼ 1 atm, T ¼ 26 C) is Z tfinal m_ N2 dt N2 VN2 ¼ 0

or VN 2 ¼

V lnð2:80Þ ð1Þð200 m3 Þð1:031Þ ¼ 206 m3 N 2

It is interesting to realize that as the N2 or air is being added, CH4 is always part of the gas mixture leaking from the room. This exhausted methane can have a flammable concentration as it now mixes with the external air. The flammable states for the exhaust stream in exterior air can be described by a succession of lines from the B–C locus to A. Figure 4.17 shows the effect of elevated pressure on the flammability of natural gas in mixtures with nitrogen and air. We see the upper limit significantly increase with

Figure 4.17

Effect of pressure on limits of flammability of natural gas–nitrogen–air mixtures at 26 C (taken from Zabetakis [5])

102

PREMIXED FLAMES

Figure 4.18

Limits of flammability of various methane–inert gas–air mixtures at 25 C and atmospheric pressure (taken from Zabetakis) [5]

pressure. Figure 4.18 shows the effect of other diluents on the flammability of CH4 , including the halogenated retardant, carbon tetrachloride ðCCl4 Þ. The optimization of a specific diluent in terms of weight or toxicity can be assessed from such a figure. It should be pointed out that such diagrams are not only used to assess the potential flammability of a mixture given an ignition source but also to assess the suppression or extinguishment of a fire (say CH4 in air) by the addition of a diluent (inert or halogenated). In Figure 4.18, the lower concentration of the ðCCl4 Þ needed to achieve the nonflammability of a stoichiometric mixture is less than the inert diluents because of the chemical inhibition effect of chloride ions on the fuel species free radicals. The free radicals are reduced and the reaction rate is effectively slowed.

4.8 Empirical Relationships for the Lower Flammability Limit Properties of the paraffin hydrocarbons ðCn H2nþ2 Þ exhibit some approximate relationships that are sometimes used to describe common fuels generally. Table 4.6 show some data [5]. The lower flammability limit (LFL) is found to be approximately proportional to the fuel stoichiometric concentration: XL ¼ 0:55Xst

ð4:51Þ

103

EMPIRICAL RELATIONSHIPS FOR THE LOWER FLAMMABILITY LIMIT Table 4.6

Flammability properties of paraffin hydrocarbons (at 25 C, 1 atm where relevant)

Fuel

M

Methane, CH4 Ethane, C2 H6 Propane, C3 H8 n-Butane, C4 H10 n-Pentane, C5 H12 n-Hexane, C6 H14 n-Heptane, C7 H16

16.0 30.1 44.1 58.1 72.2 86.2 100.

=air

hc (kJ/mole)

XL (%)

Xst (%)

XU (%)

0.55 1.04 1.52 2.01 2.49 2.98 3.46

802 1430 2030 2640 3270 3870 4460

5.0 3.0 2.1 1.8 1.4 1.2 1.1

9.5 5.6 4.0 3.1 2.5 2.2 1.9

15.0 12.4 9.5 8.4 7.8 7.4 6.7

for 25 C and 1 atm conditions. The upper limit concentration follows: pﬃﬃﬃﬃﬃﬃ XU ¼ 6:5 XL

ð4:52Þ

It is further found that the adiabatic flame temperature is approximately 1300 C for mixtures involving inert diluents at the lower flammable limit concentration. The accuracy of this approximation is illustrated in Figure 4.19 for propane in air. This approximate relationship allows us to estimate the lower limit under a variety of conditions. Consider the resultant temperature due to combustion of a given mixture. The adiabatic flame temperature ðTf;ad Þ, given by Equation (2.22) for a mixture of fuel ðXF Þ, oxygen ðXO2 Þ and inert diluent ðXD Þ originally at Tu , where all of the fuel is consumed, is N Z X i¼1 products

Tf;ad

25 C

Xi

N Z Tu X np fc ~cpi dT Xi~cpi dT ¼ XF;u h nu 25 C i¼1

ð4:53Þ

reactants

Basically, the chemical energy released in burning all of the fuel in the original (unburned) mixture ðXF;u Þ is equal to the sensible energy stored in raising the temperature

Figure 4.19 Adiabatic flame temperature and mixture concentration for propane [5,13]

104

PREMIXED FLAMES

of the products of combustion. For an approximate constant specific heat of the product and reactant mixtures, realizing for air at typical lower limits the specific heat is approximately that of nitrogen. Then Equation (4.53) becomes

Tf;ad ¼ Tu þ

fc XF;u h ~cp;u

ð4:54Þ

f c is the heat of combustion in molar where XF;u is the unburned fuel mole fraction and h units. Applying the constant adiabatic temperature approximation of 1300 C to mixtures at different initial temperatures ðTu Þ gives a relationship for the lower limit as a function of temperature: XL ðTu Þ ¼

~cp;u ð1300 C Tu Þ fc h

(4.55)

Notice that at Tu ¼ 25 C and for ~cp;u ¼ 0:030 kJ/mole K of the original unburned mixture f c 38 kJ=mole XL h

ð4:56Þ

Similar analyses can be applied to mixtures of paraffin hydrocarbons where we define Xi to be the ith fuel concentration of the fuel mixture, XLi to be the lower limit of only the single ith fuel in air and XL to be the lower limit of the fuel mixture in air. For a mixture of N fuels, the heat of combustion with respect to the fuel mixture (e.g. 10 % CH4 , 90 % C3 H8 ) is fc ¼ h

N X

fc Xi h i

ð4:57Þ

i¼1

f c is the heat of combustion for the ith fuel. Since we can approximate from where h i Equation (4.55), f c ¼ XL h f c ¼ ~cp;u ð1300 C 25 CÞ XLi h i

ð4:58Þ

Combining, ~cp;u ð1300 C 25 CÞ ¼ XL

N X i¼1

Xi

~cp;u ð1300 C 25 CÞ XLi

or NX fuels 1 Xi ¼ XL XL i i¼1

(4.59)

A QUANTITATIVE ANALYSIS

105

This is sometimes known as Le Chatelier’s law, which now gives the lower limit of the mixture of N fuels in terms of fuel concentration ðXi Þ and the respective individual lower limits ðXL;i Þ.

4.9 A Quantitative Analysis of Ignition, Propagation and Extinction The ignition, propagation and extinction of premixed combustion systems depends on detailed chemical kinetics and highly nonlinear temperature effects. Although we have qualitatively protrayed these phenomena through a series of graphs involving the energy production and heat loss terms as a function of temperature, it is difficult to quantitatively appreciate the magnitude of these phenomena. Unfortunately, it is not possible to make accurate computations without a serious effort. However, the approximate equations for flame speed and burning rate for a zeroth-order reaction offer a way to obtain order of magnitude results. We shall pursue this course. Quantitative results will depend very much on the property data we select. The following properties have been found to yield realistic results and are representative of air and fuel mixtures. u ¼ 1:1 kg=m3 cp ¼ 1:0 kJ=kg K k ¼ 0:026 W=m K ¼ k=u cp ¼ 2:4 105 m2 =s hc ¼ 40 kJ=g E ¼ 160 kJ=mole A ¼ 107 kg=m3 s Also, representative lean fuel mass fractions between the lower limit for propagation and stoichiometric conditions, range from about YF;u ¼ 0:03 to 0.05. This corresponds to adiabatic flame temperatures, from cp ðTf;ad Tu Þ ¼ YF;u hc of 1225 C to 2025 C for Tu ¼ 25 C. Using Equation (4.40), the corresponding flame speeds are computed as 0.24 and 1.7 m/s, respectively. These are realistic values according to Figure 4.8.

4.9.1

Autoignition Calculations

Consider a spherical flammable mixture of radius r ¼ 1 cm, surrounded by air at temperature T1 . The heat loss coefficient is approximated as pure conduction, h k=r ¼ 0:026=0:01 ¼ 2:2 W=m2 K

106

PREMIXED FLAMES

Figure 4.20 Autoignition calculations

From Equations (4.3) and (4.4), a solution for the minimum T1 needed to cause ignition can be found. We plot Q_ L and Q_ R as functions of T, the flammable mixture average temperature: Q_ L ¼ hð4 r 2 ÞðT T1 Þ ¼ 0:00327ðT T1 Þ W 3 6 1:92104 =T Q_ R ¼ m_ 000 W F hc ð4=3Þ r ¼ 1:67 10 e

The results are plotted in Figure 4.20. For T1 < 527 C ignition does not occur, and only stable solutions result, e.g. T1 ¼ 227 C, T ¼ 230 C. The critical tangency condition, c, is the minimum condition to initiate ignition with T1 ¼ 527 C and T ¼ 550 C. The time to reach this ignition condition can be estimated from Equations (4.11) and (4.24) and is tig tc þ tR where cp ð4=3Þ r 3 ðk=rÞ4 r 2 r2 ¼ k=cp

tc ¼

ð0:01 mÞ2 2:4 105 m2 =s ¼ 4:2 s

¼

and tR ¼ ¼

cp T1 ½E=ðRT1 Þ hc m_ 000 F ð1:1 kg=m3 Þð1:0 kJ=kg KÞð800 KÞ

f160 103 =ð8:314Þð800Þgð40 103 kJ=kgÞ107 kg=m3 s e1:92=800 ¼ 2:4 s Hence, the time ignition is about 6 s.

A QUANTITATIVE ANALYSIS

Figure 4.21

4.9.2

107

Piloted ignition calculations

Piloted ignition calculations

Consider the same spherical flammable mixure of 1 cm radius with a uniformly distributed energy source. This is an idealization of a more realistic electric arc ignition source which could be roughly 10 000 C, or a flame of roughly 2000 C, distributed over less than 1 mm. For our idealization, we plot Equations (4.26) and (4.27) as a function of the mixture temperature. Here T1 ¼ 25 C, and we determine the energy supply rate needed from the pilot. The plot in Figure 4.21 is similar to that of Figure 4.20 except that here the needed energy comes from the pilot, not the surroundings. The critical condition for ignition corresponds to a supply energy rate of 1.65 W, with anything less leading to a stable solution at an elevated temperature for as long as the fuel and oxidizer exist. At the critical condition, the mixture temperature is at 550 C, which was also found for the autoignition case. Since the heating is uniform and from within, we might estimate the ignition time as the reaction time estimated earlier at 527 C, or 2.4 s. For the uniform constant pilot at 1.65 W, this would require a minimum energy of about 4 J. Spark or electric energies are typically about 1 mJ as a result of their very high temperature over a much smaller region. The uniform pilot assumption leads to the discrepancy. Nevertheless, a relatively small amount of energy, but sufficiently high enough temperature, will induce ignition of a flammable mixture.

4.9.3

Flame propagation and extinction calculations

Once ignition has occurred in a mixture of fuel and oxidizer, propagation will continue, provided the concentrations are sufficient and no disturbance results in excessive cooling. The zeroth-order rate model is assumed to represent the lean case. Substituting the selected properties into Equation (4.43), the net release and loss curves are plotted in Figure 4.22 as a function of the flame temperature. The initial temperature of the mixture is 25 C and fuel mass fractions are 0.05 and 0.03, representative of stoichiometric and the lower limit respectively. At this lower limit, we should see that a steady solution is not possible, and the calculations should bear this out. The burning rate is evaluated at the flame temperature, and R is found from Equation (4.44) with Su at the flame temperature

108

PREMIXED FLAMES

Figure 4.22

Flame propagation calculations

also: kW=m2

3 Q_ 00R ¼ m_ 000 F R ½40 10 kJ=g 1 kJ=kg KðT 298 KÞ=YF;u Q_ 00L ¼ ð0:026 103 kW=m KÞðT 298 KÞ= R kW=m2 3 7 4 m_ 000 F ¼ 10 kg=m s expð1:92 10 =TÞ

kg=m3 s

m R ¼ 3ð2:4 105 m2 sÞ=Su " #1=2 3ð2:4 105 m2 sÞð40 103 kJ=kgÞm_ 000 F Su ¼ ð1:1 kg=m3 Þð1 kJ=kg KÞðT 298Þ

m=s:

Because the results for Q_ 00L were bound to be somewhat too large due to underestimating R , an alternative linear heat loss relationship in shown in Figure 4.22. The linear loss considers a constant reaction thickness of 1.65 mm. This shows the sensitivity of the properties and kinetic modeling parameters. Let us use the linear heat loss as it gives more sensible results. For YF;u ¼ 0:05, a distinct stable solution is found at the flame temperature of about 2000 C with a corresponding adiabatic flame temperature of 2025 C. At YF;u ¼ 0:03, extinction is suggested with a tangent intersection at about 1050 C with a corresponding adiabatic flame temperature of 1225 C.

4.9.4

Quenching diameter calculations

In this case, the loss is considered solely to the duct wall at 25 C. The net energy release is computed as above (Equation (4.43)) and the loss to the wall per unit duct crosssectional area is Q_ 00L ¼ 14:64ð0:02 103 kW=m KÞðT 298 KÞ R =D2

kW=m2

REFERENCES

Figure 4.23

109

Quenching diameter calculations

where R is evaluated as before. Figure 4.23 shows computed results for YF;u ¼ 0:05 and Tu ¼ 25 C. The diameter is varied to show the solution possibilities. The critical condition shows the quenching diameter at 0.375 mm. Combustion or stable propagation can occur at greater diameters.

References 1. Arrhenius, S., On the reaction velocity of the inversion of cane sugar by acids, Z. Phys. Chem., 1889, 4, 226–48. 2. Westbrook, C.K. and Dryer, F.L., Chemical kinetic modeling of hydrocarbon combustion, Prog. Energy Combust. Sci., 1984, 10, 1–57. 3. Semenov, N.N., Zur theorie des verbrennungsprozesses, Z. Phys. Chem., 1928, 48, 571. 4. ASTM E 659, Standard Test Method for Autoignition of Liquid Chemicals, American Society for Testing and Materials, Philadelphia, Pennsylvania, 1978. 5. Zabetakis, M.G., Flammability characteristics of combustible gases and vapors, Bureau of Mines Bulletin 627, 1965. 6. Botha, J.P. and Spalding, D.B., The laminar flame speed of propane–air mixtures with heat extraction from the flame, Proc. Royal Soc. London, Ser. A., 1954, 225, 71–96. 7. Mallard, E. and Le Chatelier, H., Recherches experimentales et theoriques sur la combustion des melanges gaseoux explosifs, Ann. Mines, 1883, 4, 379. 8. Zeldovich, Ya. B., Barenblatt, G.I., Librovich, V.B. and Makviladze, G.M., The Mathematical Theory of Combustion and Explosions, Consultants Bureau, New York, 1985, p. 268. 9. Meyer, E., A theory of flame proposition limits due to heat loss, Combustion and Flame, 1957, 1, 438–52. 10. Incroprera, F.P. and DeWitt, D.P., Fundamentals of Heat and Mass Transfer, 4th edn., John Wiley & Sons, New York, 1996, p. 444. 11. Coward, H.F. and Jones, G.W., Limits of flammability of gases and vapors, Bureau of Mines Bulletin 627, 1965. 12. Beyler, C., Flammability limits of premixed and diffusion flames, in The SFPE Handbook of Fire Protection Engineering, 2nd edn (eds P.J. Di Nenno et al.), Section 2, Chapter 9, National Fire Protection Association, Quiney, Massachusetts, 1995, p. 2–154. 13. Lewis, B. and von Elbe, G., Combustion, Flames and Explosions of Gases, 2nd edn., Academic Press, New York, 1961, p. 30.

110

PREMIXED FLAMES

Problems 4.1

The energy conservation is written for a control volume surrounding a moving premixed flame at velocity Su into a fuel–air mixture at rest. The equation is given below:

cp

d dt

ZZZ

T dV ¼ m_ 000 F R hc u Su cp ðTb Tu Þ kðTb Tu Þ= R

CV

Identify in words each of the energy terms. 4.2

The burning velocity, Su, can be determined by a technique that measures the speed of a spherical flame in a soap bubble. This process is shown below. vf Fuel + Oxygen

Tf

Tu Soap bubble, Before ignition

Soap bubble, After ignition

The flame front moves at a speed vf ¼ 6 m=s. The temperature in the burnt region is Tf ¼ 2400 K. The temperature in the unburnt region is Tu ¼ 300 K. The burned gas is at rest. Determine the burning velocity, Su. (Hint: draw a control volume around the flame and write the conservation of mass.) 4.3

A mixture of methane and air is supplied to a Bunsen burner at 25 C. The diameter of the tube is 1 cm and the mixture has an average velocity of 0.6 m/s. (a) Calculate the minimum and maximum mass flow rates of CH4 that will permit a flame for the burner. (b) For a stoichiometric mixture, the flame forms a stable conical shape at the tube exit. The laminar burning velocity for CH4 in air is 0.37 m/s. Calculate the cone half-angle, . 2

R Surface area of cone As ¼ sin

(c) What do we expect to happen if we increase the mixture flow rate while keeping the mixture at the stoichiometric condition? (d) The mixture flow velocity is reduced to 0.1 m/s while maintaining stoichiometric conditions. A flame begins to propagate back into the tube as a plane wave. Assume the propagation to be steady and then calculate the flame propagation velocity (vf).

PROBLEMS

111

Flame

φ

Bunsen Burner

D =1 cm Air

Air CH 4

4.4

Consider the ignition of a flammable mixture at the closed end of a horizontal circular tube of cross-sectional area A. The tube has a smaller opening at the other end of area Ao. The process is depicted in the figure below. The flame velocity in the tube is designated as vf. It is distinct from the ideal burning velocity, Su, which is constant in this adiabatic case. The following assumptions apply: The flame spread process in adiabatic. There is no heat transfer to the tube wall and the temperature of the burned products, Tb, and the unburned mixture, Tu, are therefore constant. Bernoulli’s equation applies at the tube exit. The contraction area ratio to the vena contracta of the exiting jet is Co. The perfect gas law applies. The pressure in the tube varies with time, but is uniform in the tube. Specific heats are constant. The reaction rate, m_ 000 F , and flame thickness, R , are related to the ideal burning speed by

u cp Su ðTb Tu Þ ¼ m_ 000 F R hc 3k Su u cp R

vb, Tb, ρb

vf

A

vu, Tu, ρu

Ao

vo

(a) Express the velocity of the flame in terms of the ideal burning speed and the velocity of the unburned mixture. (Hint: use the definition of the ideal burning speed.) (b) Express the velocity of the unburned mixture in terms of the pressure differences between the tube gas and the outside environment, p–po. (Hint: use Bernoulli.)

112

PREMIXED FLAMES (c) Derive the differential equation for the pressure in the tube. Use the nomenclature here and in the text for any other variables that enter. (Hint: use the energy equation, Equation (3.45).) Solve the energy equation and discuss the behavior of the flame velocity in the tube. (There is a limit to the speed since the exit velocity cannot exceed the speed of sound, i.e. choked flow.)

4.5

A flammable gas mixture is at a uniform concentration and at rest in a room at 25 C and 1:06 105 Pa. The properties of the gas mixture is that it has a burning velocity Su ¼ 0:5 m/s and its adiabatic flame temperature is 2130 C in air. The room is totally open (3 m 3 m)at one end where an observer turns on a light switch. This triggers a plane-wave adiabatic laminar flame propagation in the mixture that propagates over 15 m. The room processes are also adiabatic so that the unburned and burned gas temperatures are fixed.

15 m

A = 3 × 3m

(a) What is the velocity of the flame propagation as seen by the observer? (b) What is the velocity of the burned gases as they exit the room as seen by the observer? Assume the pressure is uniform in the room, but can change with time. (c) Compute the pressure rise in the compartment after 0.5 seconds. Other properties: mixture specific heats, cp ¼ 1:0 kJ/kg K, cv ¼ 0:71 kJ/kg K, p ¼ RT; 1 N=m2 ¼ 1 Pa, 1 N ¼ 1 kg m=s2 and 1 J ¼ 1 N m. 4.6

Given that the lower flammability limit of n-butane (n-C4H10) in air is 1.8 % by volume, calculate the adiabatic flame temperature at the limit. Assume the initial temperature to be 25 C. Use Table 4.5.

4.7

What is the LFL in air of a mixture of 80 % (molar) methane and 20 % propane by Le Chatelier’s rule? What is the adiabatic flame temperature of this system?

4.8

Felix Weinberg, a British combustion scientist, has shown a mixture that is 1 % by volume of methane in air and can burn if preheated to 1270 K. Although this temperature is higher than the autoignition temperature for methane in air ( 550 C), the mixture burned by pilot (‘spark’) ignition before the autoignition effects could develop. Make your own estimation of the LFL at the initial mixture temperature of 1270 K. Does it agree with the experimental results?

4.9

Calculate the adiabatic flame temperature, at constant pressure, for ethane (C2H6) in air at 25 C: (a) at the lower flammability limit (XL); (b) at the stoichiometric mixture condition (Xst). Use Tables 2.1 and 4.6.

4.10

It is reported that the adiabatic flame temperature for H2 at the lower flammability limit (LFL) in air is 700 C. From this information, estimate the LFL, in % by volume, for the hydrogen–air mixture at 25 C. Assume water is in its vapor phase within the products.

4.11

A gaseous mixture of 2 % by volume of acetone and 4 % ethanol in air is at 25 C and pressure of 1 atm.

PROBLEMS

113

Data: Acetone (C3H6O), heat of combustion: hc ¼ 1786 kJ/g mol Ethanol (C2H5OH), heat of combustion: hc ¼ 1232 kJ/g mol Atomic weights: H ¼ 1, C ¼ 12, O ¼ 16 and N ¼ 14 Specific heat, cp ; I ¼ 1 kJ/kg K, constant for each species. Find: (a) For a constant pressure reaction, calculate the partial pressure of the oxygen in the product mixture. (b) Determine the adiabatic flame temperature of this mixture. (c) Do you think this mixture is above or below the lower flammable limit? Why? (d) If this mixture was initially at 400 C, what will the resultant adiabatic flame temperature be? 4.12

Compute the lower flammable limit in % mole of acetonitrile in air at 25 C using the 1600 K adiabatic temperature rule. Acetonitrile (C2H3N(g)) burns to form hydrogen cyanide (HCN(g)), carbon dioxide and water vapor. Heat of formation in kcal/g mol Hydrogen cyanide: 32.2 Acetonitrile:

21.0

Water vapor:

57.8

Carbon dioxide:

94.1

Oxygen:

0.0

Assume constant and equal specific heats of constant pressure and constant volume of 1.2 and 1.0 kJ/kg K, respectively. 4.13

An electrical fault causes the heating of wire insulation producing fuel gases that mix instantly with air. The wire is in a narrow shaft in which air enters at a uniform velocity of 0.5 m/s at 25 C (and density 1.18 kg/m3). The cross-sectional area of the shaft is 4 cm2. The gaseous fuel generated is at 300 C and its heat of combustion is 25 kJ/g. Assume steady state conditions and constant specific heat at 1.0 J/g K. (a) What is the mass flow rate of the air into the shaft? (b) The wire begins to arc above the point of fuel generation. Draw a control volume that would allow you to compute the conditions just before ignition, and state the minimum temperature and boundary conditions needed for ignition. (c) Find the minimum mass flow rate of fuel necessary for ignition by the arc. Show the control volume used in your analysis. (d) Find the minimum fuel mass fraction needed for ignition by the arc. Show the control volume used in your analysis.

4.14

After combustion in a closed room, the temperature of the well-mixed gases achieves 350 C. There is gaseous fuel in the room left over at a mass fraction of 0.015. There is ample air

114

PREMIXED FLAMES left to burn all of this fuel. Do you consider this mixture flammable, as defined as capable of propagating a flame from an energy source? Quantitatively explain your answer. You may assume a specific heat of 1 J/g K and a heat of combustion from the fuel of 42 kJ/g.

4.15

The mole (volume) fraction of acetone (C3H6O) as a vapor in air is 6 %. This condition is uniform throughout a chamber of volume 20 m3. Air is supplied at 0.5 m3/s and the resulting mixture is withdrawn at the same rate. Assume that the mixture always has a uniform concentration of acetone for the chamber. Assume the gas density of the mixture is nearly that of pure air, 1.2 kg/m3. How long does it take to reduce the acetone mole fraction to 2 %?

4.16

A methane leak in a closed room is assumed to mix uniformly with air in the room. The room is (4 4 2:5)m high. Take the air density as 1.1 kg/m3 with an average molecular weight of 29 g/g mole. How many grams of methane must be added to make the room gases flammable? The lower and upper flammability limits of methane are 5 and 15 % by volume respectively.

4.17

How much N2 would we have to add to a mixture of 50 % by volume of methane, 40 % oxygen and 10 % nitrogen to make it nonflammable? Use Figure 4.16 and determine graphically.

4.18

Use the flammability diagram below to answer the following: (a) Is pure hydrazine flammable? Label this state A on the diagram. (b) Locate a mixture B on the diagram that is 50 % hydrazine and 45 % air. (c) If pure heptane vapor is added to mixture B find the mixture C that just makes the new mixture nonflammable. Locate C in the figure.

PROBLEMS 4.19

115

Using the methane–oxygen–nitrogen flammability diagram (Figure 4.16(b)), complete the following and show your work on the diagram: (a) Locate the mixture B of 20 % methane, 50 % oxygen and the remainder, nitrogen. Is this in the flammable range? (b) If we move methane from this mixture, find the composition just on the boundary of the flammable range.

4.20

Methane leaks from a tank in a 50 m3 sealed room. Its concentration is found to be 30 % by volume, as recorded by a combustible gas detector. The watchman runs to open the door of the room. The lighter mixture of the room gases flows out to the door at a steady rate of 50 g/s. The flammable limits are 5 and 15 % by volume for the methane in air. Assume a constant temperature at 25 C and well-mixed conditions in the room. The mixture of the room gases can be approximated at a constant molecular weight and density of 25 g/mol and 1.05 kg/m3 respectively. After the door is opened, when will the mixture in the room become flammable?

4.21

Indicate which of the following are true or false: (a) The lower flammability limit is usually about one-half of the stoichiometric fuel concentration. T___F___ (b) The lower flammability limit does not depend on the mixture temperature and pressure. T___F___ (c) The burning velocity of a natural gas flame on a stovetop burner is zero because the flame is stationary. T___F___ (d) The quenching diameter is the minimum water droplet diameter needed to extinguish a flame. T___F___ (e) The heat of combustion is related to the enthalpy of formation of all the reactant and product species. T___F___

5 Spontaneous Ignition

5.1 Introduction Spontaneous ignition or spontaneous combustion are terms applied to the process of autoignition arising from exothermiscity of the material itself. The term self-heating is also used to describe the exothermic process leading to the ignition event. As discussed in the theory of ignition, the ignition event is a criticality that results in a dramatic temperature change leading to a state of combustion. For solid materials, spontaneous ignition can manifest itself as smoldering combustion at temperatures as low as 300 C without perceptible glowing, or is visible flaming combustion with gas temperatures in excess of 1300 C. It is probably more common to apply the term spontaneous ignition to the event of flaming combustion due to self-heating. Factors that can contribute to the spontaneous ignition of solids are many: (a) bulk size: contributes to the storage of energy and interior temperature rise; (b) porosity: contributes to the diffusion of air to promote oxidation and metabolic energy release of biological organisms; (c) moisture: contributes to the growth of bacteria and other micro-organisms; (d) contamination: additives, such as unsaturated oils, can promote exothermiscity; (e) process defects: storage of heated materials, insufficient additives (such as antioxidants in synthetic polymers) or terminating a process before its chemical reactions are completed. These processes listed here are just some of the factors that can cause and promote spontaneous ignition. Common examples of spontaneous ignition have included moist haystacks, oily cotton rags and mounds of low-grade coal from mining operations. Many have experienced the Fundamentals of Fire Phenomena James G. Quintiere # 2006 John Wiley & Sons, Ltd ISBN: 0-470-09113-4

118

SPONTANEOUS IGNITION

warm interior of a mulch pile used in gardening. As with haystacks, moist woody matter will experience a biological process that produces energy. This energy can raise the temperature of the wood matter where it can oxidize or degrade and chemically produce more energy. Thus, self-heating occurs. The chemical process, in general, can be exothermic due to oxidation or due to thermal decomposition, as in ammonium potassium nitrate. Bowes [1] presents an extensive theoretical and practical review of the subject of spontaneous ignition. An excerpt is presented from this book on the propensity of spontaneous ignition and self-heating in oily rags: The amounts of material involved in the initiation of fires can be quite small. For example Taradoire [1925], investigating the self-ignition of cleaning rags impregnated with painting materials, stated that experience indicated the occurrence of self ignition in as little as 25 g of rags but, in a series of experiments, found that the best results (sic) were obtained with 75 g of cotton rags impregnated with an equal weight of mixtures of linseed oil, turpentine and liquid ‘driers’ (containing manganese resinate) and exposed to the air in cylindrical containers made of large-mesh wire gauze (dimensions not stated). Times to ignition generally lay between 1 h and 6 h, but for reasons which were not discovered, some samples, which had not ignited within these times and were considered safe, ignited after several days. Other examples giving an indication of scale have been described by Kissling [1895], who obtained self-heating to ignition, from an initial temperature of 23.5 C, in 50 g of cotton wool soaked with 100 g of linseed oil and by Gamble [1941] who reports two experiments with cotton waste soaked in boiled linseed oil (quantities not stated) and lightly packed into boxes; with the smaller of the two, a cardboard box of dimensions 10 cm 10 cm 15 cm, the temperature rose from 21 C to 226 C in 6 14 h and the cotton was charred.

We have had similar experiences in examining cotton rags impregnated with linseed oil and loosely packed in a slightly closed cardboard box about 30 cm on a side. Ignition could occur in 2 to 4 hours. It was proceeded by a noticeable odor early, then visible white ‘smokey’ vapor. It could result in a totally charred cotton residue – indicating smoldering – or into flaming ignition. On some occasions, no noticeable ignition resulted from the self-heating. This simple demonstration was utilized by a surprised newscaster on Philadelphia TV when the contents of a cardboard box with only rags burst into flames. He was exploring the alleged cause of the fire at One Meridian Plaza (1988) which led to multistory fire spread, an unusable 38 story building, three firefighter fatalities and damage in civil liabilities of up to $400 million. A less expensive accident attributed to wood fiberboard in 1950 brought the subject of spontaneous ignition to the forefront of study. This is vividly described in the opening of an article by Mitchelle [2] from the NFPA Quarterly, ‘New light on self-ignition’: One carload of wood fiberboard burned in transit and eight carloads burned in a warehouse 15 days after shipment, causing a $2,609,000 loss. Tests to determine the susceptibility of wood fiberboard to self-heating revealed that the ambient temperature to cause ignition varied inversely with the size of the specimen, ranging from 603 F for a 18-inch cube to 252 F for an octagonal 12-in prism. Pronounced self-heating was evident when the surrounding air was held at 147 F. An untoward event sometimes points the way to useful discoveries. A fire in an Army Warehouse, June 1950, pointed the way to the subject of this discussion. The story briefly is: on June 2, nine carloads of insulation fiberboard were shipped from a factory in the South, bound for upstate New York. During the afternoon of June 9, in the course of switching, one carload was discovered to be afire and left in the railyard. The other eight cars were sent on a few miles farther to their destination, where they were unloaded into a warehouse during the period June 12 to 15. This lot of eight carloads of fiberboard was stacked in one pile

THEORY OF SPONTANEOUS IGNITION Table 5.1

119

Ignition temperatures for wood fiberboard [2]

Characteristic dimension (in)

Ignition temperature ( F)

1/8 12 22

603 252 228

measuring more than 24,000 cubic feet. In the forenoon of June 17, two days afterward, fire was discovered which destroyed the warehouse and its contents. The monetary loss was reported to have been $2,609,000.

It is evident from the excerpt of Mitchell’s article that the temperature necessary to cause ignition of the fiberboard depends on the size of the wood. Table 5.1 gives data from Mitchell [2]. The implication is that a stack of the wood of 24 000 ft3 (or about 29 ft as a characteristic length) led to ignition in the warehouse after two days in a New York June environment at roughly 75 F. Such events are not uncommon, and have occurred for many types of stored processed materials.

5.2 Theory of Spontaneous Ignition Although many factors (chemical, biological, physical) can be important variables, we shall restrict our attention to the effects of size and heating conditions for the problem. The chemical or biological effects will be accounted for by a single Arrhenius zerothorder rate relationship. The material or medium considered will be isotropic and undergo no change in phase with respect to its thermal properties, taken as constant. Thus, no direct accounting for the role and transport of oxygen and fuel gases will be explicit in the following theory. If we could include these effects it would make our analysis more complex, but we aim to remain simple and recognize that the thermal effects are most relevant. Since size is very important; we will consider a one-dimensional slab subject to convective heating. The geometry and heating condition can be modified resulting in a new problem to solve, but the strategy will not change. Hence, one case will be sufficient to illustrate the process of using a theory to assess the prospect of spontaneous ignition. Later we will generalize it to other cases. We follow the analysis of Frank-Kamenetskii [3] of a slab of half-thickness, ro , heated by convection with a constant convective heat transfer coefficient, h, from an ambient of T1 . The initial temperature is Ti T1 ; however, we consider no solution over time. We only examine the steady state solution, and look for conditions where it is not valid. If we return to the analysis for autoignition, under a uniform temperature state (see the Semenov model in Section 4.3) we saw that a critical state exists that was just on the fringe of valid steady solutions. Physically, this means that as the self-heating proceeds, there is a state of relatively low temperature where a steady condition is sustained. This is like the warm bag of mulch where the interior is a slightly higher temperature than the ambient. The exothermiscity is exactly balanced by the heat conducted away from the interior. However, under some critical condition of size ðro Þ or ambient heating (h and T1 ), we might leave the content world of steady state and a dynamic condition will

120

SPONTANEOUS IGNITION

Figure 5.1

Theoretical model

ensue. This is the phenomenon of thermal runaway or thermal explosion, which means that we will move to a state of combustion – smoldering or flaming. Let us derive the governing equation for steady state conduction in a one-dimensional slab of conductivity, k, with internal heating described by an energy release rate per unit volume 000 Q_ ¼ ðAhc Þe E=ðRTÞ

ð5:1Þ

where ðAhc Þ and E are empirical properties. Consider a small element, x, of the slab (Figure 5.1) by the conservation of energy for the control volume surrounding the element 000

00

00

0 ¼ Q_ x þ q_ x q_ xþx

ð5:2Þ

By Fourier’s law of conduction, 00

q_ x ¼ k

dT dx

ð5:3Þ

and by Taylor’s expansion theorem, 00

00

q_ xþx ¼ q_ x þ

d 00 ðq_ Þx þ OðxÞ2 dx x

ð5:4Þ

By substitution in Equation (5.2), dividing by x and letting x ! 0, it follows that d 00 000 0 ¼ Q_ ðq_ x Þ OðxÞ dx or d dT k þ ðAhc Þe E=ðRTÞ ¼ 0 dx dx

ð5:5Þ

THEORY OF SPONTANEOUS IGNITION

121

The boundary conditions are dT ¼0 dx

x ¼ 0;

by symmetry

ð5:6Þ

and x ¼ ro ;

k

dT ¼ hðT T1 Þ dx

ð5:7Þ

by applying the conservation of energy to the surface. It is useful to rearrange these equations into dimensionless form as was done in Section 4.3, Equation (4.9): T T1 E ¼ RT1 T1

(5.8)

e E=ðRTÞ e e E=ðRT1 Þ

ð5:9Þ

where again we approximate

anticipating that the critical temperature will be close to T1 . With ¼ x=ro

ð5:10Þ

d2 þ e ¼ 0 d2

ð5:11Þ

the problem becomes

with d=d ¼ 0 at ¼ 0 and d=d þ Bi ¼ 0 at ¼ 1. The dimensionless parameters are: (a) the Damkohler number, : ¼

E RT1

ro2 ðAhc Þe E=ðRT1 Þ kT1

(5.12)

which represents the ratio of chemical energy to thermal conduction, or the ratio of thermal diffusion ðro2 =Þ to chemical reaction time from Equations (4.23) and (4.24), and (b) the Biot number, Bi: Bi ¼

hro k

ð5:13Þ

122

SPONTANEOUS IGNITION

the ratio of convection to conduction heat transfer in the solid. If h is very large for high flow conditions, Bi ! 1. Mathematically, this reduces to the special case of Tðro Þ ¼ T1 or ð1Þ ¼ 0. As in the solution to the simpler Semenov problem (see section 4.3), we expect that as increases (increasing the exothermiscity over the material’s ability to lose heat) a critical c is reached. At this point, an unsteady solution must hold and we have no valid solution to Equation (5.11). For a given geometric shape of the solid and a given heating condition, there will be a unique c that allows this ‘ignition’ event to occur. For the slab problem described, we see that since Bi is an independent parameter, then c must be a function of Bi. For the special case of a constant temperature boundary ð ¼ 1; ¼ 0Þ, FrankKamenetskii [3] gives the solution to Equation (5.11) as e ¼

a h pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ i cosh b a=2 2

ð5:14Þ

where a and b are constants of integration. From Equation (5.6), it can be shown that b ¼ 0, and from ð1Þ ¼ 0, it follows that a must satisfy a ¼ cosh2

a 2

ð5:15Þ

The maximum temperature at the center of the slab can be expressed from Equation (5.14) as em ¼ a

ð5:16aÞ

m ¼ lnðaÞ ¼ fðÞ

ð5:16bÞ

or

From Equation (5.15), a is an implicit function of and m is a function only of . A solution is displayed in Figure 5.2. Steady physical solutions are given on the lower

Figure 5.2 Solution for a slab at a constant surface temperature

THEORY OF SPONTANEOUS IGNITION

Figure 5.3

123

c As a function of Bi for a sphere, cylinder and slab (from Bowes [1])

branch; i.e. for small values of < c ¼ 0:88, the slab is able to maintain a state in which the chemical energy is conducted to the surface at T1 . From Equation (5.12), can increase as T1 increases or as ro increases. If exceeds 0.88 for any reason, a steady solution is not possible, and a thermal runaway occurs leading to ignition, and finally to another possible steady state at a new temperature corresponding to combustion. The upper branch of the curve has mathematically possible steady solutions but is believed to be unstable, and thus not relevant. Therefore, to prevent spontaneous ignition of this slab, we must be assured that its Damkohler number is less than 0.88. Another configuration or heating condition will have another solution and a different critical Damkohler number, c . The critical dimensionless temperature corresponding to c ¼ 0:88 is approximately m; c ¼ 1:18. As found for Equation (4.19), Tm; c T1 RT1 T1 4 ¼ ð1:18Þ T1 E 10 as a typical result for E 30 kcal/mole. This confirms that Tm; c at ignition is a small departure from T1 , and T1 can be regarded as the ignition temperature corresponding to a given ro . It is not a unique property of the material! Figure 5.3 and Table 5.2 give solutions for c for convective heating of a sphere, cylinder and slab. As the degrees of freedom for heat conduction increase, it is less likely to achieve ignition as c increases from a slab (1), to a cylinder (2), to a sphere (3). Also as Bi ! 0, the material becomes perfectly adiabatic, and only an unsteady solution is

Table 5.2

Critical Damkohler numbers for Bi ! 1 (medium) c

Semenov, ro ¼ V=S Slab, ro is half-width Infinite cylinder, ro is radius Sphere, ro is radius Cube, ro is half-side

e

1

¼ 0:368 0.88 2.0 3.32 2.52

124

SPONTANEOUS IGNITION

possible, c ! 0. For Bi ! 1, the constant surface temperature case, c for a cube of half side, ro , is 2.52, and for a rectangular box of half dimensions ro ; ‘o ; and so is "

2 2 # ro ro c ¼ 0:88 1 þ þ ‘o so

5.3 Experimental Methods In order to assess spontaneous ignition propensity for a material practically, data must be developed. According to the model presented, these data would consist of ðE=RÞ and ðAhc =kÞ from Equation (5.12) for . If Bi is finite, we must also know k separately. To simplify our discussion, let us consider only the constant surface temperature case ðBi ! 1Þ. Experiments are conducted in a convective oven with good circulation (to make h large) and at a controlled and uniform oven temperature, T1 (Figure 5.4). Data of cubical configurations of sawdust, with and without an oil additive, are shown in Table 5.3. The temperatures presented have been determined by trial and error to find a sufficient T1 (T1; 2 Figure 5.4) to cause ignition. Ignition should be observed to determine whether it is flaming or smoldering. The center temperature, Tm , is used as the indicator of thermal runaway. A surface temperature should also be recorded to examine whether it is consistent with our supposition that Ts ¼ T1 for the convective oven. If this is not true, then h must be determined for the oven. While the experiment is simple, care must be taken to properly secure the thermocouples to the material. Also, patience is needed since it can take many hours before ignition occurs. If a sample changes and melts, the data can be invalid. This is particularly true since small oven-size samples require relatively high temperatures for ignition, and such temperatures are likely to exceed the melting point of common polymers. Hence, the test conditions must be as close as possible to the conditions of investigation; however, this may not be fully possible to achieve. Figure 5.5 gives data from Hill [4] for a cotton material used in the buffing of aluminum for aircraft surfaces. This was tested because fires were periodically occurring in a hopper where the waste cotton was discarded. It can be seen from these data that only a small oven temperature change is needed when we are near the critical conditions, i.e. c .

Figure 5.4 Experimental method

EXPERIMENTAL METHODS

125

Table 5.3 Critical oven temperatures for cubes of sawdust with and without oil (from Bowes [1]) Cube size 2ro (mm) 25.4 25.4 51 51 76 76 152 152 303 303 910 910 a

Oil content (%) 0 11.1 0 11.1 0 11.1 0 11.1 0 11.1 0 11.1

Ignition temperature ( C) 212 208 185 167 173 146 152a 116 135 99 109a 65

Calculated.

Figure 5.5

Example of a thermal response for a cotton-based material used in an aluminum polishing process (from Hill [4])

126

SPONTANEOUS IGNITION

The needed properties can be derived from data illustrated in Table 5.3. To derive these data, it is convenient to rearrange Equation (5.12) as 2 c T1 E 1 ln ¼P ð5:17Þ R T1 ro2 where P ln

E Ahc R k

Since the data are for cubes, c ¼ 2:52 and we plot " # T1 ðKÞ 2 1 ln c versus ro ðmÞ T1 ðKÞ in Figure 5.6. This gives information on the bulk kinetic properties of the sawdust. The fact that the data are linear in ð1=T1 Þ suggests that the theory we are using can, at least, explain these experimental results. From the slope of the lines, E=R is 9:02 103 K and 13:8 103 K for oil and no added oil to the sawdust respectively. For R ¼ 8:314 J=mole K, we obtain the activation energies of Esd ¼ 115 kJ=mole and Esd; oil ¼ 75 kJ=mole showing that the oil-impregnated sawdust can react more easily. If we approximate, for illustrative purposes, a thermal conductivity of the sawdust as 0.1 W/m K, then E Ahc 39:5; ln ¼ 49:3; R k

with oil no oil

or Ahc ¼

1:6 1012 W=m3 ; 1:9 1016 W=m3 ;

with oil no oil

Figure 5.6 Bulk kinetics of sawdust

TIME FOR SPONTANEOUS IGNITION

127

These results give some appreciation for the kinetic properties. They are ‘effective’ properties since the actual mechanism of the exothermic reaction is more complex than a zeroth-order Arrhenius model. However, it is seen to be quite satisfactory for these data. The practical question is the accuracy of extrapolating these results to large cubes, or other sawdust configurations and conditions. Nevertheless, the procedure described does give a powerful analytical tool, when combined with good data, to make a quantitative judgment about the propensity of a product to reach spontaneous ignition. Let us consider a practical application. Example 5.1 Animal feedstock, a distillery byproduct, is to be stored in a silo, a rectangular facility 12 m high and 3 m 3 m square. For an ambient temperature of 20 C do we expect a problem? What would be the critical ambient temperature sufficient to cause spontaneous ignition in the silo? The following oven data was obtained for cubes of side sð¼ 2ro Þ from feedstock samples. s(mm) 50 75 150 300

Critical oven temperature ( C 3 C) 158 140 114 85

Solution Using the procedure just outlined with c ¼ 2:52 for a cube, we can show that P ¼ 39:88 in units of ro in m and E=R ¼ 8404 K. Substituting these values into Equation (5.12) yields r 2 ðmÞ 8404 exp ¼ 2:081 1017 o2 T1 ðKÞ T1 ðKÞ as a general equation for the Damkohler number corresponding to any ro and T1 for the animal feedstock. This can now be applied to the silo configuration. For the rectangular silo, we compute the corresponding critical Damkohler number, c ðro Þ: " 2 2 # 1:5 1:5 þ c ¼ 0:88 1 þ ¼ 1:80 6 1:5 Note that this c is defined in terms of ro ¼ 1:5 m. Calculating for ro ¼ 1:5 m and T1 ¼ 20 C from our feedstock equation gives ¼ 1:9. Since the for this condition is greater than the critical value of 1.8, we will have spontaneous ignition. For ¼ c ¼ 1:80 and ro ¼ 1:5 m, we solve for T1 to find that the critical ambient temperature is 19.5 C. Any value T1 below 19.5 C will be safe.

5.4 Time for Spontaneous Ignition The prospect of successfully estimating the likelihood of spontaneous ignition depends on our ability to know c for our configuration and heating condition, and the bulk kinetic

128

SPONTANEOUS IGNITION

properties ðE=RÞ and ðAhc =kÞ of the material. Since the kinetics are very sensitive to temperature, any uncertainties can be magnified by its nonlinear effects. Therefore, it may be pushing our hopes to consider an estimate of the time for spontaneous condition. However, since the self-heating period can be hours or even days before spontaneous ignition occurs, it is useful to examine its dependence. The Frank-Kamenetskii model for a given geometry is too complex to achieve analytical results, so we return to the Semenov model of Section 4.3, Equation (4.13): d ¼ e d

ð5:18Þ

This deals with purely convective heating, but defining by Equation (5.12), we take h ¼ k=ro and ro ¼ V=S of the Semenov model. Equation (5.18) approximates the constant surface temperature case. Also, here the initial temperature is taken as T1 . If we allow no convective cooling, the adiabatic equation becomes d ¼ e d

ð5:19Þ

Equation (5.19) can also be obtained from the unsteady counterpart of Equation (5.5). If internal conduction is taken as zero in Equation (5.5), this adiabatic equation arises. Equation (5.19) also becomes the basis of an alternative method for determining the bulk kinetic properties. The method is called the ‘adiabatic furnace’ used by Gross and Robertson [5]. The furnace is controlled to make its temperature equal to the surface temperature of the material, thus producing an adiabatic boundary condition. This method relies on the measurement of the time for ignition by varying either ro or T1, the initial temperature. Now let us return to determining an estimate for the ignition time with the sample initially at T1 . (There may be a thermal conduction time needed to bring the sample to T1 ; this is ignored.) The time to ignite under adiabatic conditions will give a lower bound on our estimate. Equation (5.19) can be integrated to give Z 0

d ¼

Z

e d

0

or ¼ 1 e

ð5:20Þ

where is the time (dimensionless) to achieve . We assume that the dimensionless critical temperature for ignition is always a constant value. For the nongeometrydependent Semenov case (from Equation (4.17)), c ¼ 1 and from the constant surface temperature slab case, m; c ¼ 1:18

TIME FOR SPONTANEOUS IGNITION

129

Hence, selecting c 1 appears to be a reasonable universal estimate. Then from Equation (5.20), ig ¼ 1 e 1 ¼ 0:632 or

ig 0:632=

ð5:21aÞ

In dimensional terms, this adiabatic ignition time is 0:632ro2

ð5:21bÞ

tig; ad 0:632tR

(5.21c)

tig; ad If we substitute Equation (5.12) for ,

where tR is the chemical reaction time given in Equation (4.24). The characteristic time chosen for making time dimensionless was tc

ro2

which is the time for thermal diffusion through the medium of the effective radius, ro . Again, this shows that is the ratio of these times: ¼

tc tR

Applying Equations (5.21) to the adiabatic time corresponding to the critical Damkohler number, and realizing for a three-dimensional pile of effective radius, ro ; c 3 (e.g. c ¼ 3:32 for a sphere for Bi ! 1), then we estimate a typical ignition time at ¼ c 3 of tig; c ¼ 0:21ro2 = To estimate the time to ignite when > c , from Equations (5.21) and (5.22), 2 c r c tig tig; c ¼ 0:21 o

ð5:22Þ

ð5:23Þ

Equation (5.23) is still based on the adiabatic system. A better, but still approximate, result can be produced from the nonadiabatic Semenov model where d ¼ e d

ð5:24Þ

130

SPONTANEOUS IGNITION

from Equation (4.13). We know at ignition that 1 and c ¼ e 1 ; or e c 1 Then Equation (5.24), moving toward ignition, might be approximated as d 1 d c

ð5:25Þ

Integrating from the initial state to ignition, Z

1

d

o

1 ig c

ð5:26aÞ

or tig

ðro2 =Þ ð=c 1Þ

ð5:26bÞ

Beever [6] reports an approximation of Boddington et al. for the nonadiabatic case, as

tig M h

ro2 =

ð=c 1Þ1=2

i

ð5:27Þ

where M is approximately 1.6. Good accuracy is claimed for ð=c Þ < 3. For a given size, ro , where the ambient temperature is greater than the critical temperature, the ratio of the Damkohler numbers is eðT1 =T1;c 1Þ½E=ðRT1 Þ ¼ c ðT1 =T1; c Þ2

ð5:28Þ

A typical value of E=ðRT1 Þ is about 30, and with an ambient temperature of 10 % above the critical temperature, =c ¼ 16:5. Thus, the time for ignition is greatly reduced as the temperature exceeds its critical value.

References 1. Bowes, P.C., Self-Heating: Evaluating and Controlling the Hazards, Elsevier, The Netherlands, 1984. 2. Mitchelle, N.D., New light on self-ignition, NFPA Quarterly, October 1951, 139. 3. Frank-Kamenetskii, D.A., Diffusion and Heat in Chemical Kinetics, 2nd edn., Plenum Press, New York, 1969. 4. Hill, S.M., Investigating materials from fire using a test method for spontaneous ignition: case studies, MS Thesis, Department of Fire and Protection Engineering, University of Maryland, College Park, Maryland, 1997.

PROBLEMS

131

5. Gross, D. and Robertson, A.F., Self-ignition temperatures of materials from kinetic-reaction data, J. Research National Bureau of Standards, 1958, 61, 413–7. 6. Beever, P.F., Self-heating and spontaneous combustion, in The SFPE Handbook of Fire Protection Engineering, 2nd edn (eds P.J. DiNenno et al.), Section 2, National Fire Protection Association, Quincy, Massachusetts, 1995, p. 2-186. 7. Gray, B.F. and Halliburton, B., The thermal decomposition of hydrated calcium hypochlorite (UN 2880), Fire Safety J., 2000, 35, 233–239.

Problems 5.1

In June 1950 an Army Warehouse was destroyed by fire with a $2 609 000 loss. It occurred two days after a shipment of fiberboard arrived in which a freight car of the shipment had experienced a fire. The remaining fiberboard was stored in a cubical pile of 24 000 cubic feet. Spontaneous ignition was suspected. Data for the fiberboard was developed (Gross and Robertson [5]): ¼ 0:25 g/cm3 cp ¼ 0:33 cal/g K k ¼ 0:000 12 cal/s cm K E ¼ 25:7 kcal/mol Ahc ¼ 1:97 109 cal/s cm3 Determine whether an ambient temperature of 50 C could cause the fire. If not consider 100 C.

5.2

A cubical stack of natural foam rubber pillows was stored in a basement near a furnace. That part of the basement could achieve 60 C. A fire occurs in the furnace area and destroyed the building. The night watchman is held for arson. However, an enterprising investigator suspects spontaneous ignition. The stack was 4 meters on a side. The investigator obtains the following data for the foam: ¼ 0:108 g/cm3 cp ¼ 0:50 cal/g K k ¼ 0:000 096 cal/s cm K E ¼ 27:6 kcal/mol Ahc ¼ 7:48 1010 cal/s cm3 Does the watchman have a plausible defense? Show calculations to support your view.

5.3

Calculate the radius of a spherical pile of cotton gauze saturated with cottonseed oil to cause ignition in an environment with an air temperature (Ta ) of 35 C and 100 C. Assume perfect heat transfer between the gauze surface and the air. The gauze was found to follow the Frank-Kamenetskii ignition model, i.e.

k

d2 T ¼ Ahc e E=ðRTÞ dx2

132

SPONTANEOUS IGNITION Small-scale data yield: k ¼ 0:000 11 cal/s cm K E ¼ 24 100 cal/mol Ahc ¼ 2:42 1011 cal/s cm3 Note: R ¼ 8:31 J/mol K

5.4

A fire disaster costing $67 million occurred in Texas City, Texas, on the SS Grandcamp (16 April 1987) due to spontaneous ignition of stored fertilizer in the ship’s hold. A release of steam from an engine leak caused the atmosphere of the ammonium nitrate fertilizer to be exposed to temperatures of 100 C. Ammonium nitrate (NH4 NO3 ) decomposes exothermically releasing 378 kJ/g mol. Its rate of decomposition can be described by the Arrhenius equation: 000

m_ ¼ ð6 1013 Þe 170 kJ=g mol=ðRTÞ in kg=m3 s where T is temperature in K Density, ¼ 1750 kg/m3 Thermal conductivity, k ¼ 0:1256 W/m K and Universal gas constant, R ¼ 8:3144 J/g mol K (a) What is the minimum cubical size of the fertilizer under this steam atmosphere to achieve self-ignition? (b) Explain what happens to the remaining stored fertilizer after self-ignition of this cube. 5.5

A flat material of thickness ‘ is placed on a hot plate of controlled temperature Tb . The material is energetic and exothermic with a heat of combustion of hc and its reaction is governed by zeroth-order kinetics, Ae E=ðRTÞ , the mass loss rate per unit volume. Notation is as used in the text. The differential equation governing the process to ignition is given as

c

@T @2T ¼ Ae E=ðRTÞ hc þ k 2 @t @x

where x is the coordinate measured from the hot plate’s surface. Let ¼ x=‘ and ¼ T=Tb . Convection occurs at the free boundary with a heat transfer coefficient hc and the surrounding temperature at T1 . Invoke the approximation that 1 is small (see Equation (4.8)), and use the new dimensionless temperature variable ¼ ðE=ðRT1 Þð 1Þ. Write the governing equation and the boundary and initial conditions in terms of the dimensionless variables : ; and , 5.6

A polishing cloth residue is bundled into a cube of 1 m on the half-side dimension. It is stored in an environment of 35 C and the convective heat transfer is very good between the air and the bundle. Experiments show that the cloth residue has the following properties: Thermal conductivity, k ¼ 0:036 W/m K Density, ¼ 68:8 kg/m3

PROBLEMS

133

Specific heat, c ¼ 670 J/kg K Heat of combustion, hc ¼ 15 kJ/g Kinetic properties: A ¼ 9:98 104 kg/m3 s E=R ¼ 8202 K (a) Show that it is possible for this bundle to spontaneously ignite. (b) Compute an estimate for the time for ignition to occur. 5.7

A material is put on a hot plate. It has a cross-sectional area, S, and thickness, ro . The power output of the plate is P. The plate and edges of the material are perfectly insulated.

The material can be chemically exothermic, and follows the theory:

d2 þ e ¼ 0 d2 where ¼ x=ro ;

¼

T T1 T1

E RT1

;

is the Damkohler number

Boundary conditions for the material: At the hot plate: k dT dx ¼ P=S At the ambient: k dT dx ¼ hc ðT T1 Þ with T1 the ambient temperature and hc the convective coefficient. From the boundary conditions, show the dimensionless parameters that the critical Damkohler number will depend on. 5.8

Hydrated calcium hypochlorite exothermically decomposes to release oxygen. The heat of decomposition is relatively small, but the resulting limited increase in temperature coupled with the increase in oxygen concentration can cause the ignition of adjacent items. Ignition has occurred in its storage within polyethylene (PE) kegs. Consider the keg to be an equicylinder – a cylinder of radius ro and height 2ro . The critical Damkohler number and critical temperature are found for this geometry to be c ¼ 2:76 and c ¼ 1:78 for a large Bi number. Data collected by Gray and Halliburton [7] are given below for an equicylinder geometry.

134

SPONTANEOUS IGNITION Radius, ro (m)

Critical temperature ( C)

0.00925 0.01875 0.02875 0.055 0.075 0.105 0.175 0.175

149.5, 154.2 135.5, 141.5 126.5, 128.5 120.5, 123.0 102.5, 101.0 87.5, 90.6 64.0 60.1 (in PE keg)

(a) Plot the data as in Figure 5.6. Show that there are two reactions. Find the Arrhenius parameters and the temperature range over which they apply. The thermal conductivity can be taken as 0.147 W/m K. (b) For an equicylinder exposed to an ambient temperature of 45 C, determine the critical radius for ignition, and the center temperature at the onset of ignition. 5.9

The parameter entered into the solutions for spontaneous ignition. (a) What is the name associated with this dimensionless parameter? (b) Explain its significance in the Semenov and Frank-Kamenetskii models with respect to whether ignition is possible. (c) For a given medium, spontaneous ignition depends on ambient temperature as well as other factors. Name one more. (d) Once conditions are sufficient for the occurrence of spontaneous ignition, it will always occur very quickly. True or false?

5.10

The time for spontaneous ignition to occur will decrease as the Damkohler number _____increases or _____decreases with respect to its critical value.

5.11

List three properties or parameters that play a significant role in causing spontaneous ignition of solid commodities.

6 Ignition of Liquids

6.1 Introduction The ignition of a liquid fuel is no different from the processes described for autoignition and piloted ignition of a mixture of gaseous fuel, oxidizer and a diluent, with one exception. The liquid fuel must first be evaporated to a sufficient extent to permit ignition of the gaseous fuel. In this regard, the fuel concentration must at least be at the lower flammable limit for piloted ignition, and within specific limits for autoignition. The liquid temperature corresponding to the temperatures at which a sufficient amount of fuel is evaporated for either case is dependent on the measurement apparatus. For piloted ignition, the position of the ignitor (spark) and the mixing of the evaporated liquid fuel with air is critical. For autoignition, both mixing, chamber size and heat loss can make a difference (see Figure 6.1). Temperature and fuel concentration play a role, as illustrated in Figure 6.2 for piloted ignition in air. The fuel can only be ignited in a concentration range of XL to XU , and a sufficient temperature must exist in this region, corresponding to piloted and autoignition respectively. These profiles of XF and T depend on the thermodynamic, transport properties and flow conditions. We will see that the heat of vaporization (hfg ) is an important thermodynamic property that controls the evaporation rate and hence the local concentration. The smallest heating condition to allow for piloted ignition would be with a spark at the surface when XF;s ¼ XL , the lower flammable limit.

6.2 Flashpoint The liquid temperature (TL ) corresponding to XL is measured for practical purposes in two apparatuses known as either the ‘closed’ or ‘open’ cup flashpoint test, e.g. ASTM D56 and D1310. These are illustrated in Figure 6.3. The surface concentration (Xs ) will be shown to be a unique function of temperature for a pure liquid fuel. This temperature is known as the saturation temperature, denoting the state of thermodynamic equilibrium Fundamentals of Fire Phenomena James G. Quintiere # 2006 John Wiley & Sons, Ltd ISBN: 0-470-09113-4

136

IGNITION OF LIQUIDS

Figure 6.1 Autoignition of an evaporated fuel in a spherical vessel of heated air (taken from Setchkin [1])

Figure 6.2 Dynamics of piloted ignition for a liquid fuel in air under convective heating from the atmosphere

between a liquid and its vapor. Theoretically, the saturation temperature corresponding to XL is the lower limit temperature, TL . Because the spark location is above the liquid surface and the two configurations affect the profile shape of XF above the liquid, the ignition temperature corresponding to the measurement in the bulk liquid can vary between the open and closed cup tests. Typically, the closed cup might insure a lower T at ignition than the open cup. The minimum liquid temperature at which a flame is seen at the spark is called the flashpoint. This temperature may need to be raised to achieved sustained burning for the liquid, and this increased temperature is called the firepoint. The flashpoint corresponds to the propagation of a premixed flame, while the firepoint corresponds to the sustaining of a diffusion flame (Figure 6.4). In these test apparatuses, the liquid fuel is heated or cooled to a fixed temperature and the spark is applied until the

Figure 6.3

Fuel concentration differences in the closed and open cup flashpoint tests

DYNAMICS OF EVAPORATION

Figure 6.4

137

Flashpoint and firepoint, liquid fuel at a uniform temperature

lowest temperature for ignition is found. These ‘flashpoints’ are commonly reported for the surrounding air at 25 C and 1 atm pressure. They will vary for atmospheres other than air since XL changes. However, XL tends to range from about 1 to 8 % for many common conditions. Thus, a relatively small amount of fuel must be evaporated to permit ignition. It is useful to put in perspective the range of temperature conditions during the combustion of a liquid fuel. In increasing order: 1. TL ¼ flashpoint or the saturation temperature corresponding to the lower flammable limit. 2. Tb ¼ boiling point corresponding to the saturation pressure of 1 atm. 3. Ta ¼ autoignition temperature corresponding to the temperature that a mixture of fuel and air can self-ignite. Usually this is measured at or near a stoichiometric mixture. 4. Tf;ad ¼ adiabatic flame temperature corresponding to the maximum achievable temperature after combustion. This is usually reported for a stoichiometric mixture of fuel in air. Table 6.1 gives these temperature data along with the heat of vaporization and heat of combustion, two significant variables. The former controls the evaporation process and the latter is significant for sustained burning. To have sustained burning it is necessary that hc > hfg ; hfg is usually a weak function of temperature and is usually taken as constant for a fuel at temperatures relevant to the combustion process. Table 6.2 gives some additional relevant properties.

6.3 Dynamics of Evaporation Let us examine methanol. Its flashpoint temperature is 12 to 16 C (285–289 K) or, say, 15 C. If this is in an open cup, then the concentration near the surface is XL ¼ 6:7 %. Performed under normal room temperatures of, say, 25 C, the temperature profile would be as in Figure 6.2. This must be the case because heat must be added from the air to cause this evaporated fuel vapor at the surface. This decrease in temperature of an evaporating surface below its environment is sometimes referred to as evaporative cooling. If the convective heat transfer coefficient, typical of natural convection, is,

138 Table 6.1

IGNITION OF LIQUIDS Approximate combustion properties of liquid fuels in air from various sources [2,3]

Fuel

Formula

Methane Propane n-Butane n-Hexane n-Heptane n-Octane n-Decane Kerosene Benzene Toluene Naphthalene Methanol Ethanol n-Butanol Formaldehyde Acetone Gasoline

CH4 C3 H8 C4 H10 C6 H14 C7 H16 C8 H18 C10 H22 C14 H30 C6 H6 C7 H8 C10 H8 CH3 OH C2 H5 OH C4 H9 OH CH2 O C3 H6 O —

a b

TL (K) Closed Open — — — 251 269 286 317 322 262 277 352 285 286 302 366 255 228

— 169 213 247 — — — — — 280 361 289 295 316 — 264 —

Tb (K)

Ta (K)

Tf ;ad a (K)

XL (%)

hfg (kJ/g)

hbc (kJ/g)

111 231 273 342 371 398 447 505 353 383 491 337 351 390 370 329 306

910 723 561 498 — 479 474 533 771 753 799 658 636 616 703 738 644

2226 2334 2270 2273 2274 2275 2277 — 2342 2344 — — — — — 2121 —

5.3 2.2 1.9 1.2 1.2 0.8 0.6 0.6 1.2 1.3 0.9 6.7 3.3 11.3 7.0 2.6 1.4

0.59 0.43 0.39 0.35 0.32 0.30 0.28 0.29 0.39 0.36 0.32 1.10 0.84 0.62 0.83 0.52 0.34

50.2 46.4 45.9 45.1 44.9 44.8 44.6 44.0 40.6 41.0 40.3 20.8 27.8 36.1 18.7 29.1 44.1

Based on stoichiometric combustion in air. Based on water and fuel in the gaseous state.

say, 5 W/m2 K, then the heat transfer is 00 q_ ¼ 5 W=m2 K ð25 15Þ K ¼ 50 W=m2

A corresponding mass flux of fuel leaving the surface and moving by diffusion to the pure air with YF;1 ¼ 0 would be given approximately as 00

m_ F ¼ hm ðYF ð0Þ YF;1 Þ Table 6.2

Approximate physical properties of liquid fuels from various sources [4] Liquid propertiesa

Fuel

Formula

ðkg=m3 Þ

cp ðJ=g KÞ

Methane Propane n-Butane n-Hexane n-Heptane n-Octane Benzene Toluene Naphthalene Methanol Ethanol n-Butanol Acetone

CH4 C3H8 C4H10 C6H14 C7H16 C8H18 C6H6 C7H8 C10 H8 CH3OH C2H5OH C4H9OH C3H6O

— — — 527 676 715 894 896 1160 790 790 810 790

— 2.4 2.3 2.2 2.2 2.2 1.7 1.7 1.2 2.5 2.4 2.3 2.2

a

ð6:1Þ

At normal atmospheric temperature and pressure.

Vapor Properties ðkg=m3 Þ 0.72(0 C) 2.0(0 C) 2.62(0 C) — — — — — — — 1.5 (100 C) — —

cp ðJ=g KÞ 2.2 1.7 1.7 1.7 1.7 1.7 1.1 1.5 1.4 1.4 1.7 1.7 1.3

DYNAMICS OF EVAPORATION

Figure 6.5

139

p–v Liquid–vapor diagram

where hm is a convective mass transfer coefficient. YF ð0Þ theoretically corresponds to the lower flammable limit, or Mmeth 32 YF ð0Þ ¼ ð0:067Þ ¼ 0:073 ¼ 0:067 29 Mmix Up to now we have presented this example without any regard for consistency, i.e. satisfying thermodynamic and conservation principles. This fuel mass flux must exactly 00 equal the mass flux evaporated, which must depend on q_ and hfg . Furthermore, the concentration at the surface where fuel vapor and liquid coexist must satisfy thermodynamic equilibrium of the saturated state. This latter fact is consistent with the overall approximation that local thermodynamic equilibrium applies during this evaporation process. The conditions that apply for the saturated liquid–vapor states can be illustrated with a typical p–v, or ð1=Þ, diagram for the liquid–vapor phase of a pure substance, as shown in Figure 6.5. The saturated liquid states and vapor states are given by the locus of the f and g curves respectively, with the critical point at the peak. A line of constant temperature T is sketched, and shows that the saturation temperature is a function of pressure only, Tsat ðpÞ or psat ðTÞ. In the vapor regime, at near normal atmospheric pressures the perfect gas laws can be used as an acceptable approximation, pv ¼ ðR=MÞT, where R=M is the specific gas constant for the gas of molecular weight M. Furthermore, for a mixture of perfect gases in equilibrium with the liquid fuel, the following holds for the partial pressure of the fuel vapor in the mixture: pF ¼ psat ðTÞ ¼ XF p

(6.2)

from Dalton’s law, where p is the mixture pressure. Since theoretically XF ¼ XL at the condition of piloted ignition, the surface temperature T must satisfy the thermodynamic state equation pF ¼ psat ðTÞ for the fuel. We will derive an approximate relationship for this state equation in Section 6.4.

140

IGNITION OF LIQUIDS

Figure 6.6 Equilibrium and flash evaporation processes illustrated for water

Here we are talking about evaporation under thermodynamic equilibrium. We can also have evaporation under nonequilibrium conditions. For example, if the pressure of a liquid is suddenly dropped below its saturation pressure, flash evaporation will occur. The resulting vapor will be at the boiling point or saturation temperature corresponding to the new pressure, but the bulk of the original liquid will remain (out of equilibrium) at the former higher temperature. Eventually, all of the liquid will become vapor at the lower pressure. The distinction between flash evaporation and equilibrium evaporation is illustrated in Figure 6.6 for water. An interesting relevant accident scenario is the ‘explosion’ of a high-pressure tank of fuel under heating conditions. This is illustrated in Figure 6.7 for a propane rail tank car whose escaping gases from its safety valve have been ignited. The overturned tank is heated, and its liquid and vapor segregate due to gravity, as shown. The tank metal shell adjacent to the vapor region cannot be cooled as easily compared to that by boiling in the

Figure 6.7

BLEVE Illustration

CLAUSIUS–CLAPEYRON EQUATION

141

liquid region. Eventually the heated metal’s structural properties weaken due to a temperature rise, and a localized rupture occurs in the vapor region. At this instant (before vapor expansion can occur), we have a constant volume (and constant mass) process for the liquid–vapor propane mixture. This is illustrated by 1 to 2 on Figure 6.7. Since this is a sudden event, the contents do not have a chance to cool to the new equilibrium state. The contents of the tank now seek the external pressure, p2 ¼ 1 atm. By the perfect gas law we have a considerable increase in vapor volume, R T1 v3 ¼ ð6:3Þ > vg;2 > vg;1 M p2 which in turn causes a more dramatic rupture of the tank. In this case, lack of thermodynamic equilibrium for the gas phase in achieving v3 , instead of the smaller vg;2 , exacerbates the problem. If the leakage at the rupture cannot accommodate this change in vapor volume, a pressure rise will cause a more catastrophic rupture. This phenomenon is commonly called a BLEVE (boiling liquid expanding vapor explosion). Indeed, the liquid–vapor system will expand to meet the atmospheric pressure, p2 ¼ 1 atm, at the temperature T1 and all of the mass (liquid and vapor) will reach state 3. The boiling point of the liquid propane is 231 K (42 C), which is well below normal atmospheric temperatures. Not only will the liquid receive significant heat transfer, but the new equilibrium state 3 does not allow any liquid to exist. This constant pressure process 2 ! 3 is referred to as a flash evaporation of the liquid. The liquid state cannot be maintained for atmospheric conditions of approximately 25 C and 1 atm since it is above the boiling point of 42 C. The actual process from 1 to 3 is not in two steps as shown, since it depends on the rupture process. The new volume based on v3 and the mass of the fuel can be considerable. This sudden release of a ‘ball’ of gaseous fuel leads to a fireball culminating in a mushroom shaped fire that has a relatively short lifetime.

6.4 Clausius–Clapeyron Equation The Clausius–Clapeyron equation provides a relationship between the thermodynamic properties for the relationship psat ¼ psat ðTÞ for a pure substance involving two-phase equilibrium. In its derivation it incorporates the Gibbs function ðGÞ, named after the nineteenth century scientist, Willard Gibbs. The Gibbs function per unit mass is defined as G=m ¼ g h Ts

ð6:4Þ

where s is entropy per unit mass. For a pure substance at rest with no electrical or magnetic effects, the first law of thermodynamics for a closed system is expressed for a differential change as du ¼ q w

ð6:5Þ

where both q and w are incremental heat added and work done by the substance per unit mass. For this simple system (see Section 2.4.1) w ¼ p dv

ð6:6Þ

142

IGNITION OF LIQUIDS

if no viscous or frictional effects are present, and q ¼ T ds

ð6:7Þ

if the process is reversible according to the second law of thermodynamics. Combining Equations (6.5) to (6.7) gives a relationship only between thermodynamic (equilibrium) properties for this (simple) substance, du ¼ T ds p dv

ð6:8Þ

Equation (6.8) is a general state equation for the pure substance. From the definition of enthalpy, h ¼ u þ pv, Equation (6.7) can be expressed in an alternative form:

then or

dh ¼ du þ p dv þ v dp dh ¼ T ds þ v dp

ð6:8aÞ

dg ¼ dh T ds s dT dg ¼ v dp s dT

ð6:8bÞ

Let us apply Equation (6.8) to the two-phase liquid–vapor equilibrium requirement for a pure substance, namely p ¼ pðTÞ only. This applies to the mixed-phase region under the ‘dome’ in Figure 6.5. In that region along a p-constant line, we must also have T constant. Then for all state changes along this horizontal line, under the p–v dome, dg ¼ 0 from Equation (6.8b). The pure end states must then have equal Gibbs functions: gf ðTÞ ¼ gg ðTÞ

ð6:9Þ

Consequently, it is permissible to equate the derivatives, as these are continuous functions, dgf dgg ¼ dT dT

ð6:10Þ

From Equation (6.8b), by division with dT or by a formal limit operation, we recognize that in this two-phase region, dg dp ¼v s dT dT

ð6:11Þ

From Equation (6.8a) in this two-phase region along a p ¼ constant line, Tds ¼ dh However, T is also constant here, so integrating between the two limit extremes gives Tðsg sf Þ ¼ hg hf

ð6:12Þ

CLAUSIUS–CLAPEYRON EQUATION

143

Substitution of Equation (6.11) into Equation (6.10) gives vf

dp dp sf ¼ vg sg dT dT

ð6:13Þ

By combining Equations (6.12) and (6.13), it follows that dp sg sf ¼ ¼ dT vg vf

hg hf 1 T vg vf

(6.14)

This is known as the Clausius–Clapeyron equation. It is a state relationship that allows the determination of the saturation condition p ¼ pðTÞ at which the vapor and liquid are in equilibrium at a pressure corresponding to a given temperature. An important extension of this relationship is its application to the evaporation of liquids into a given atmosphere, such as air. Consider the evaporation of water into air (XO2 ¼ 0:21 and XN2 ¼ 0:79). Suppose the air is at 21 C. If the water is in thermal equilibrium with the air, also at 21 C, its vapor at the surface must have a vapor pressure of 0.0247 atm (from standard Steam Tables). However, if the water and the air are at the same temperature, no further heat transfer can occur. Therefore no evaporation can take place. We know this cannot be true. As discussed, in the phenomenon of ‘evaporative cooling’, the surface of the liquid water will have to drop in temperature until a new equilibrium pðTÞ can satisfy the conservation laws, i.e. Mass evaporation rate ¼ diffusion transport rate to the air Heat transfer rate ¼ energy required to vaporize Let us say that this new surface equilibrium temperature of the water is 15.5 C; then the vapor pressure is 0.0174 atm. At the surface of the water, not only does water vapor exist, but so must the original gas mixture components of air, O2 and N2 . By Dalton’s law the molar concentration of the water vapor is XH2 O ð15:5 CÞ ¼ 0:0174 and the new values for O2 and N2 are XO2 ¼ 0:21ð1 0:0174Þ ¼ 0:2063 XN2 ¼ 0:79ð1 0:0174Þ ¼ 0:776 For this value of the water vapor concentration at the surface, XH2 O ð15:5 CÞ ¼ 0:0174, the relative humidity () at this point for the atmosphere at 21 C is 0:0174=0:0247 or 70.4 %. In general, relative humidity is given as

¼

pH2 O ðT1 Þ 100 % pg ðT1 Þ

(6.15)

144

IGNITION OF LIQUIDS

Figure 6.8

Relative humidity and dew point

where pH2 O is the partial pressure of water vapor in the atmosphere and pg is the water saturation pressure at the temperature of the atmosphere, T1 . The dew point ðTdp Þ is the saturation temperature corresonding to pH2 O ðT1 Þ. The relationship of these temperatures are illustrated in Figure 6.8. In other words, if the atmosphere at a given relative humidity (RH) is suddenly reduced in temperature to Tdp , the new state becomes 100 % relative humidity. Any further cooling will cause the water vapor in the atmosphere to condense. This process causes the dew on the grass in early morning as the blades of grass have undergone radiative cooling to the black night sky and their temperature has dropped below the dew point of the ‘moist’ atmosphere. By assuming that a perfect gas law applies for the vapor, including the saturation state, it can be seen that

100

¼

vg pH2 O RT1 =vH2 O MH2 O H O ¼ ¼ ¼ 2 pg RT1 =vg MH2 O vH 2 O g

ð6:16Þ

Thus at 100 % relative humidity the air atmosphere contains the most mass of water vapor possible for the temperature T1 . No further evaporation of liquid water is possible. The introduction of the perfect gas law to the Clausius–Clapeyron equation (Equation (6.14)) allows us to obtain a more direct approximation to p ¼ pðTÞ in the saturation region. We use the following: 1. pvg ¼ ðR=M ÞT. 2. hg hf hfg , the heat of vaporization at temperature T. 3. vg vf ; vg vf vg , except near the critical point. Substituting into Equation (6.14), we obtain hfg ðTÞ dp ¼ 2 dT T R=ðMpÞ

ð6:17Þ

If we assume that hfg ðTÞ is independent of T, reasonable under expected temperatures, then we can integrate this equation. The condition selected to evaluate the constant of

CLAUSIUS–CLAPEYRON EQUATION

145

integration is p ¼ 1 atm, T ¼ Tb , which is the definition of the boiling point usually reported (see Table 6.1). With Mg the molecular weight of the evaporated component, Z

p

hfg dp ¼ p R=M g p1 ¼ 1 atm

Z

T

Tb

dT T2

or hfg Mg T¼T p ln ¼ p1 RT T¼Tb or Xg ¼

p ¼ eðhfg Mg =RÞð1=T1=Tb Þ p1

ð6:18Þ

where p1 ¼ 1 atm for the normally defined boiling point, or in general is the mixture pressure and Tb is the saturation temperature where p ¼ p1 . Notice that this equation has a similar character to the Arrhenius form, with hfg Mg playing the role of an activation energy (kJ/mole). Example 6.1 Estimate the minimum piloted ignition temperature for methanol in air which is at 25 C and 1 atm. Solution

Use the data in Table 6.1:

Tb ¼ 337 K ¼ 64 C hfg ¼ 1:10 kJ/g Mg ¼ 12 þ 4 þ 16 ¼ 32 g/mole R ¼ 8:315 J/mole K ^ hfg ¼ ð1:1Þð32Þ ¼ 35:2 kJ/mole The minimum temperature for piloted ignition is given by XL ¼ Xg ðTL Þ. From Equation (6.18), 35:2 kJ=mole 1 1 0:067 ¼ exp 8:315 103 kJ=mole K TL 337 K Solving gives TL ¼ 277 K ¼ 4 C This should be compared to the measured open or closed cups of 285 and 289 K. We expect the computed value to be less, since the ignitor is located above the surface and mixing affects the ignitor concentration of fuel. Thus, a higher liquid surface temperature is needed to achieve XL at the ignitor.

146

IGNITION OF LIQUIDS

6.5 Evaporation Rates Let us consider a case of steady evaporation. We will assume a one-dimensional transport of heat in the liquid whose bulk temperature is maintained at the atmospheric temperature, T1 . This would apply to a deep pool of liquid with no edge or container effects. The process is shown in Figure 6.9. We select a differential control volume between x and x þ dx, moving with a surface velocity ððdx0 =dtÞ iÞ. Our coordinate system is selected with respect to the moving, regressing, evaporating liquid surface. Although the control volume moves, the liquid velocity is zero, with respect to a stationary observer, since no circulation is considered in the contained liquid. We apply the conservation of mass to the control volume: d dx0 dx0 ð dxÞ þ ðx þ dxÞ 0 i ðþiÞ þ ðxÞ 0 i ðiÞ ¼ 0 dt dt dt where is the mean density between x and x þ dx. As we formerly divide by dx and take the limit as dx ! 0, with ðx þ dxÞ expressed in terms of x by Taylor’s expansion theorem ðx þ dxÞ ðxÞ þ

@ dx @x

we obtain @ @ v0 ¼0 @t x @x t

ð6:19Þ

where v0 ¼ dx0 =dt. Since the density is constant in the liquid, this equation is automatically satisfied.

Figure 6.9

Liquid evaporation

147

EVAPORATION RATES

It is more significant to take a control volume between x ¼ 0 (in the gas) and any x in the liquid. For this control volume (CVII), Z d x 00 dx þ ðxÞ½0 ðv0 iÞ ðþiÞ þ m_ g ¼ 0 dt 0 00

where m_ g is the mean flux of vapor evaporated at the surface. Since is constant 00

v0 ¼ m_ g

(6.20)

The energy equation can be derived in a similar fashion for CVI. The liquid is considered to be at the same pressure as the atmosphere and therefore and Equation (3.48) applies: cp

h 00 ixþdx d ðT dxÞ þ cp v0 ½TðxÞ Tðx þ dxÞ ¼ q_ k dt x

00

where q_ k is the heat flux in the x direction. Expanding and taking the limit as dx ! 0, 00

cp

@T @T @ q_ cp v0 ¼ k @t @x @x

ð6:21Þ

Considering Fourier’s law for heat conduction, 00

q_ k ¼ k

@T @x

ð6:22Þ

Then we combine Equations (6.21) and (6.22) to obtain cp

@T @T @2T cp v0 ¼k 2 @t @x @x

(6.23)

for k ¼ constant. At x ¼ 0, we can employ a control volume (CVIII) just surrounding the surface. For this control volume, the conservation of mass is Equation (6.20). The conservation of energy becomes at x ¼ 0 :

00

00

00

m_ g hg v0 hf ¼ q_ q_ k

ð6:24Þ

00

where q_ is the net incident heat flux to the surface. Note that there is no volumetric term in the equation since the volume is zero for this ‘thin’ control volume. From Equations (6.20) and (6.22), we obtain at x ¼ 0 :

@T m_ g hfg ¼ q_ þ k @x x¼0 00

00

ð6:25Þ

148

IGNITION OF LIQUIDS

We also know at x ¼ 0, T ¼ Ts that the saturation temperature at pg ð0Þ or Xg ð0Þ Xs ¼ pg ð0Þ=p1 , in terms of molar concentrations. At x ! 1, T ¼ T1 for a deep pool of liquid. In summary, for steady evaporation, Equation (6.23) becomes v0

dT d2 T ¼ 2 dx dx

(6.26)

The thermal diffusivity ¼ k=cp. We have a second-order differential equation requiring two boundary conditions: x ¼ 0; T ¼ Ts x ! 1; T ¼ T1

ð6:27aÞ ð6:27bÞ

The evaporation rate is also unknown. However, for pure convective heating, we can write 00

q_ ¼ hðT1 Ts Þ

ð6:28Þ

The convective heat transfer coefficient may be approximated as that due to heat transfer without the presence of mass transfer. This assumption is acceptable when the evaporation rate is small, such as drying in normal air, and for conditions of piloted ignition, since XL is typically small. Mass transfer due to diffusion is still present and can be approximated by 00

m_ g ¼

h ðYs Y1 Þ cg

(6.29)

where the convective mass transfer coefficient is given as h=cg (cg is the specific heat at constant pressure for the gas mixture) with Y1 as the mass fraction of the transferred phase (liquid ! gas) in the atmosphere. For the evaporation of water into dry air, Y1 ¼ 0. If we consider the unknowns in the above equations, we count four: Ts ; Ys ; m_ g

and

q_

00

The differential equation can be solved to give T ¼ Tðx; Ts ; T1 ; v0 = Þ. The four unknowns require four equations: Equations (6.26), (6.28) and (6.29), together with the Clausius–Clapeyron equation in the form of Ys ðTs Þ ¼

Mg ðMg hfg =RÞð1=Ts 1=Tb Þ e M

(6.30)

where M is the molecular weight of the gas mixture (M 29 for air mixtures). Furthermore, it can be shown that where the mass transfer rate is small, Ts will be

EVAPORATION RATES

149

only several tens of degrees, at most, below T1. Hence we might avoid solving explicity for Ts by approximating Ys ðTs Þ Ys ðT1 Þ

ð6:31Þ

Let us complete the problem by solving Equation (6.26). Let ¼

dT dx

Therefore, d v0 ¼ dx Integrating once gives ¼ c1 eðv0 = Þx and integrating again gives ðv0 = Þx T ¼ c1 þ c2 e v0

ð6:32Þ

By the boundary conditions of Equation (6.27), c2 ¼ T1 and T s T1

¼ c1 v0

Then, the solution can be expressed as T T1 ¼ ðTs T1 Þeðv0 = Þx Substituting into Equation (6.25) gives k

v dT 0 ¼ kðTs T1 Þ dx x¼0 ¼ cp v0 ðTs T1 Þ 00

¼ m_ g cp ðTs T1 Þ

(6.33)

150

IGNITION OF LIQUIDS

(cp is the specific heat of the liquid) or 00 00 q_ ¼ m_ g hfg þ cp ðTs T1 Þ

(6.34)

where hfg þ cp ðTs T1 Þ is called the heat of gasification since it is the total energy needed to evaporate a unit mass of liquid originally at temperature T1 . It is usually given by the symbol L and is composed of the heat of vaporization plus the ‘sensible energy’ needed to bring the liquid from its original temperature T1 to its evaporation temperature Ts . The special case of a dry atmosphere, under the one-dimensional steady assumptions with a low mass transfer flux, can be written as (from Equations (6.29) and (6.30)) 00

m_ g ¼

Mg Mg hfg =ðRTb Þ Mg hfg =ðRTs Þ h e e cg M

ð6:35Þ

This shows that surface evaporation into a dry atmosphere has the form of an Arrhenius rule for a zeroth-order chemical reaction. Example 6.2 Estimate the mass flux evaporated for methanol in dry air at 25 C and 1 atm. Assume natural convection conditions apply at the liquid–vapor surface with h ¼ 8 W/m2 K. Solution

By Equation (6.35) and Example 6.1, we first assume Ts 25 C: 00

m_ g ¼

8 W=m2 K 32 1:0 J=g K 29 ½ð32Þð1:10 kJ=gÞ=ð8:315 103 kJ=mole KÞð1=298 K 1=337 KÞ e

00

m_ g ¼ 1:7 g=m2 s where we have approximated the specific heat of the gas mixture as 1.0 J/g K. This is a typical fuel mass flux at ignition, and at extinction also; normally 1–5 g/m2 s. The approximate heat flux needed to make this happen (for Ts T1 ¼ 25 C) is 00

00

q_ ¼ m_ g hfg ¼ ð1:7 g=m2 sÞð1:10 kJ=gÞ ¼ 1:88 kW=m2 From Equation (6.27), we can now estimate Ts as 00

q_ Ts ¼ T1 h

1:88 kW=m2 8 103 kW=m2 K Ts ¼ 25 C 235 ¼ 210 C ¼ 25 C

EVAPORATION RATES

Figure 6.10

151

Example 6.2, solution for Ts

which obviously does not satisfy our assumption that Ts ¼ T1 ¼ 25 C! This calls for a more accurate solution. We do this as follows: equate Equations (6.28) 00 and (6.34) as q_ =h; and substitute Equation (6.35): T 1 Ts ¼

Mg ðMg hfg =RÞð1=Ts 1=Tb Þ e hfg cp ðT1 Ts Þ cg M

This equation requires an iterative solution for Ts. We plot the LHS and RHS as functions of Ts for the following substituted equations: 298 K Ts ¼

ð32 g=moleÞ=ð29 g=moleÞ 103 kJ=gK 3

eð32 g=moleÞð1:10 kJ=gÞ=ð8:315 10 kJ=moleÞð1=Ts 1=337 KÞ

1:10 kJ=g ð2:37 103 kJ=gKÞ ð298 K Ts Þ giving an intersection (see Figure 6.10) for Ts 264 K ¼ 9 C Equation (6.18) could now be used to assess whether the methanol vapor exceeds its LFL. At this temperature, by Equations (6.28) and (6.34), and with cp ¼ 2:37 J=g K for liquid methanol, 00

m_ g ¼ ¼

hðT1 Ts Þ hfg cp ðT1 Ts Þ 8 103 kW=m2 K ð298 264Þ K ½1:10 ð2:37 103 Þð298 264Þ kJ=g

00

m_ g ¼ 0:26 g=m2 s Hence this accurate solution gives a big change in the surface temperature of the methanol, but the order of magnitude of the mass flux at ignition is about the same, Oð1 g=m2 sÞ.

152

IGNITION OF LIQUIDS

Let us examine some more details of this methanol problem. The regression rate of the surface is, for a liquid density of 0:79 g=cm3 , 00

v0 ¼

m_ g

¼

0:26 g=m2 s 0:79 g=cm3 106 cm3 =m3

or v0 ¼ 3:3 107 m=s From Equation (6.33), the temperature profile in the liquid pool is T 25 C ¼ eðv0 = Þx ð9 25Þ C where, using k ¼ 0:5 W=m K for the liquid, v0 ð0:26 g=m2 sÞð2:37 J=g KÞ 00 ¼ 1:23 m1 ¼ v0 cp =k ¼ m_ g cp =k ¼ 0:5 W=m K The thermal penetration depth of the cooled liquid can be approximated as eð1:23Þp ¼ 0:01 for Tðp Þ ¼ 24:7 C or p ¼ 3:7 m. Hence this pool of liquid must maintain a depth of roughly 4 m while evaporating in order for our semi-infinite model to apply. If it is a thin pool then the temperature at the bottom of the container or ground will have a significant effect on the result. Suppose the bottom temperature of the liquid is maintained at 25 C for a thin pool. Let us consider this case where the bottom of the pool is maintained at 25 C. For the pool case, the temperature is higher in the liquid methanol as depth increases. This is likely to create a recirculating flow due to buoyancy. This flow was ignored in developing Equation (6.33); only pure conduction was considered. For a finite thickness pool with its back face maintained at a higher temperature than the surface, recirculation is likely. Let us treat this as an effective heat transfer coefficient, hL , between the pool bottom and surface temperatures. For purely convective heating, conservation of energy at the liquid surface is 00

m_ g hfg ¼ hðT1 Ts Þ þ hL ðTbot Ts Þ:

ð6:36Þ

Substituting Equation (6.35) and taking Tbot ¼ T1 gives an implicit equation for Ts : hfg Mg ðhfg Mg =RÞð1=Ts 1=Tb Þ h T1 Ts ¼ e ð6:37Þ h þ hL cg M This gives the surface temperature for a liquid evaporating pool convectively heated from above and below.

EVAPORATION RATES

153

Example 6.3, solution for Ts

Figure 6.11

Example 6.3 Consider Example 6.2 for a shallow pool of methanol with its bottom surface maintained at 25 C. Assume that natural convection occurs in the liquid with an effective convective heat transfer coefficient in the liquid taken as 10 W=m2 K. Find the surface temperature, surface vapor mass fraction and the evaporation flux for this pool. Solution

From Equation (6.37), we substitute, for known quantities, 298 K Ts ¼

8 8 þ 10

32 29

1100 J=g 4240:9ð1=Ts 1=337 KÞ e 1 J=g K

The LHS and RHS are plotted in Figure 6.11, yielding Ts ¼ 272 K or 1 C. From Equation (6.30), Ys ¼

32 4240:9ð1=2721=337Þ ¼ 0:054 e 29

The mass flux evaporated into dry air is, by Equation (6.29), 00

m_ g ¼

h 8 W=m2 K ð0:054Þ Ys ¼ cg 1 J=g K

or 00

m_ g ¼ 0:44 g=m2 s

154

IGNITION OF LIQUIDS

Figure 6.12 Methanol evaporation into air at 25 C under natural convection (not drawn to scale)

These results are close to the theoretical flammable limit conditions at the surface of 00 Ts ¼ 277 K, Ys ¼ 0:074 and m_ g ¼ 0:58 g=m2 s (see Example 6.1). The thermal dynamics of Examples 6.2 and 6.3 for methanol are sketched in Figure 6.12.

References 1. Setchkin, N.P., Self-ignition temperature of combustible liquids, J. Research, National Bureau of Standards, 1954, 53, 49–66. 2. Kanury, A.M., Ignition of liquid fuels, in The SFPE Handbook of Fire Protection Engineering, 2nd edn, (eds P.J. DiNenno et al.), Section 2, National Fire Protection Association, Quincy, Massachusetts, 1995, pp. 2-164 to 2-165. 3. Turns, S.R., An Introduction to Combustion, 2nd edn, McGraw-Hill, Boston, Massachusetts, 2000. 4. Bolz, R.E. and Tuve, G.L. (eds), Handbook of Tables for Applied Engineering Science, 2nd edn, CRC Press, Cleveland, Ohio, 1973.

Problems 6.1

A runner is in a long race on a cloudy day and is perspiring. Will the joggerss’ skin temperature be _____higher or _____lower than the air temperature? Check one.

6.2

Why is the vapor pressure of a liquid fuel important in assessing the cause of the ignition of the fuel?

6.3

Under purely convective evaporation from gas phase heating, the liquid surface temperature is ______less than or____ greater than the ambient. Check the blank that fits.

6.4

At its boiling point, what is the entropy change of benzene from a liquid to vapor per gram? Use Table 6.1.

PROBLEMS 6.5

155

Calculate the vapor pressure of the following pure liquids at 0 C. Use the Clausius– Claperyon equation and Table 6.1. (a) n-Octane (b) Methanol (c) Acetone

6.6

Calculate the range of temperatures within which the vapor–air mixture above the liquid surface in a can of n-hexane at atmospheric pressure will be flammable. Data are found in Table 4.5. Calculate the range of ambient pressures within which the vapor/air mixture above the liquid surface in a can of n-decane (n-C10H22) will be flammable at 25 C.

6.7

Using data in Table 6.1, calculate the closed cup flashpoint of n-octane. Compare your results with the value given in Table 6.1.

6.8

Calculate the temperature at which the vapor pressure of n-decane corresponds to a stoichiometric vapor–air mixture. Compare your result with the value quoted for the firepoint of n-decane in Table 6.1.

6.9

A liquid spill of benzene, C6H6, falls on a sun-heated roadway, and is heated to 40 C. Safety crews are concerned about sparks from vehicles that occur at the surface of the benzene. Show by calculation that the concern of the crew is either correct or not. Use the following data as needed: Tboil ¼ 353 K hc ¼ 40.7 kJ/g

M ¼ 78 Xlower ¼ 1.2 %

hfg ¼ 432 kJ/kg Xupper ¼ 7.1 %

R ¼ 8.314 J/mole K 6.10

A bundle of used cleaning rags and its flames uniformly impinge on a vat of naphthalene (C10H8). The liquid naphthalene is filled to the brim in a hemispherical vat of 1 m radius. The flame heats the thin metal vat with a uniform temperature of 500 C and a convective heat transfer coefficient of 20 W/m2 K, and the liquid in the vat circulates so that it can be taken at a uniform temperature. Heat transfer from the surface of the liquid can be neglected, and the metal vat has negligible mass. A faulty switch at the top surface of the vat continually admits an electric arc. Using the properties for the naphthalene, determine the time when the liquid will ignite. In this process you need to compute the temperature of the liquid that will support ignition. The initial temperature is 35 C.

156

IGNITION OF LIQUIDS Properties: Liquid density, ¼ 1160 kg/m3 Liquid specific heat, c ¼ 1.2 J/g K Heat of vaporization, hfg ¼ 0.32 kJ/g Lower flammability limit, XL ¼ 0.9 % Boiling point, Tb ¼ 491 K Gas constant, R ¼ 8.315 J/mole K Surface area of the sphere ¼ 4r2 Volume of the sphere ¼ (4/3)r3 (a) Find the temperature required for ignition. (b) Find the time for ignition.

6.11

n-Octane spills on a hot pavement during a summer day. The pavement is 40 C and heats the octane to this temperature. The wind temperature is 33 C and the pressure is 1 atm. Use Table 6.1. Data: Ambient conditions are at a temperature of 33 C and pressure of 760 mmHg. Density of air is 1.2 kg/m3 and has a specific heat of 1.04 J/g K. Octane properties: Boiling temperature is 125.6 C Heat of vaporization is 0.305 kJ/g Specific heat (liquid) is 2.20 J/g K Specific heat (gas) is 1.67 J/g K Density (liquid) is 705 kg/m3 Lower flammability limit in air is 0.95 % Upper flammability limit in air is 3.20 % (a) Determine the molar concentration of the octane vapor at the surface of the liquid spill. (b) If a spark is placed just at the surface, will the spill ignite? Explain the reason for your answer. (c) Assume a linear distribution of octane vapor over the boundary layer where it is 4 cm thick. Determine the vertical region over the 4 cm thick boundary layer where the mixture is flammable.

6.12

This problem could be boring or it could be exciting. You are to determine, by theoretical calculations, the drying characteristics of a wet washcloth. A cloth is first saturated with water, then weighed and hung to dry for a 40 minutes, after which it is weighed again. The room temperature and humidity are recorded. Theoretically, compute the following, stating any assumptions and specifications in your analysis: (a) The heat and mass transfer coefficients for the cloth suspended in the vertical orientation. (b) The average evaporation rate per unit area and the drying time, assuming it continues to hang in the same position. (c) The surface temperature and the mass fraction of water vapor at the surface of the cloth.

PROBLEMS

157

Data:

6.13

Room temperature: 23 C

Room humidity: 29.5%

Washcloth mass: 55.48 g

Dimensions: 33 cm 32.5 cm 3 cm

Initial wet washcloth mass: 135.8 g

Time: 11:25 a.m.

Final mass: 122.6 g

Time: 12:05 p.m.

The US Attorney General authorizes the use of ‘nonpyrotechnic tear gas’ to force cult members from a building under siege. She is assured the tear gas is not flammable. The teargas (CS) is delivered in a droplet of methylene chloride (dichloromethane, CH2Cl2) as an aerosol. Twenty canisters of the tear gas are simultaneously delivered, each containing 0.5 liters of methylene chloride. The properties of the CH2Cl2 are listed below: Properties of CH2Cl2: Liquid specific heat ¼ 2 J/g K Boiling point ¼ 40 C Liquid density ¼ 1.33 g/cm3

Heat of vaporization ¼ 7527 cal/g mole Lower flammability limit ¼ 12 % (molar) (a) Compute the flashpoint of the liquid CH2Cl2. (b) Compute the evaporation rate per unit area of the CH2Cl2 once released by the breaking canisters in the building. The air temperature is 28 C, and no previous tear gas was introduced. Assume the spill can be considered as a deep pool, and the heat transfer coefficient between the floor and air is 10 W/m2 K. (c) What is the maximum concentration of CH2Cl2 gas in the building assuming it is well mixed. The building is 60 ft 30 ft 8 ft high, and has eight broken windows 3 ft 5 ft high. Each canister makes a 1 m diameter spill, and the air flow rate into the building can 1=2 be taken as 0.05 A0H0 in kg/s, where A0 is the total area of the windows and H0 is the height of a window.

158 6.14

IGNITION OF LIQUIDS A burglar in the night enters the back room of a jewelry store where the valuables are kept. The room is at 32 C and is 3.5 m 3.5 m 2.5 m high. He enters through an open window 0.8 m 0.8 m high. The colder outside temperature will allow air to flow into the room through the window based on the formula

pﬃﬃﬃﬃﬃﬃ m_ air ðg=sÞ ¼ 110A0 H0 where A0 is the area of the window in m2 and H0 is its height in m. The burglar trips and breaks his flashlight and also a large bottle of methanol that spills into a holding tank of 1.8 m diameter. He then uses a cigarette lighter to see. Use property data from the tables in Chapter 6; the gas specific heat is constant at 1.2 J/g K, the density for the gas in the room can be considered constant at 1.18 kg/m3 and the heat transfer coefficient at the methanol surface is 15 W/m2 K. (a) If the burglar immediately drops the lit lighter into the methanol tank, explain by calculations what happens. You need to compute the concentration of the methanol. (b) What is the initial evaporation rate of the methanol in g/s? (c) If the burglar does not drop his lit lighter, but continues to use it to see, will an ignition occur and when? Explain your results by calculations. Assume the composition of the evaporated methanol is uniformly mixed into the room, and further assume that the methanol surface temperature remains constant and can be taken for this estimate at the temperature when no methanol vapor was in the room. 6.15

Tung oil is fully absorbed into a cotton cloth, 0.5 cm thick, and suspended in dry air at a uniform temperature. The properties of the tung oil and wet cloth are given below: Heat of combustion

40 kJ/g

Heat of vaporization

43.0 kJ/mole tung oil

Lower flammability limit of tung oil

1 % (by volume)

Boiling point of tung oil

159 C

Thermal conductivity of wet cloth

0.15 W/m K

Activation energy, E

75 kJ/mole

Pre-exponential factor, A

0.38 108 g/m3 s

Gas constant, R

8.315 J/mole K

(a) Ignition occurs when a small flame comes close to the cloth. What is the minimum cloth surface temperature to allow flaming ignition? (b) If the cloth were exposed to air at this minimum temperature for (a), could it have led to ignition without the flame coming close to the cloth? Explain in quantitative terms.

7 Ignition of Solids

7.1 Introduction Up to now we have considered ignition phenomena associated with gas mixtures for auto and piloted cases (Chapter 4), the conditions needed for spontaneous ignition of bulk products (Chapter 5) and the conditions needed for the ignition of liquids. The time for ignition to occur has not been carefully studied. In fire scenarios, the role of ignition for typical solid commodities and furnishings plays an important role in fire growth. In particular, the time for materials to ignite can help to explain or anticipate the time sequence of events in a fire (time-line). Indeed, we will see (Chapter 8) that the time to ignite for solids is inversely related to the speed of flame spread. In this chapter, we will confine our discussion to the ignition of solids, but liquids can be included as well provided they are reasonably stagnant. A hybrid case of a liquid fuel absorbed by a porous solid might combine the properties of both the liquid and the solid in controlling the process. To understand the factors controlling the ignition of a solid fuel, we consider several steps. These steps are depicted in Figure 7.1. Step 1 The solid is heated, raising its temperature to produce pyrolysis products that contain gaseous fuel. This decomposition to produce gaseous fuel can be idealized by an Arrhenius type reaction 000

m_ F ¼ As eEs =ðRTÞ

ð7:1Þ

where As and Es are the properties of the solid. As with pure liquid evaporation (Equation (6.35)) this is a very nonlinear function of temperature in which a critical temperature, Tpy , might exist that allows a significant fuel vapor concentration to make ignition

Fundamentals of Fire Phenomena James G. Quintiere # 2006 John Wiley & Sons, Ltd ISBN: 0-470-09113-4

160

IGNITION OF SOLIDS

Figure 7.1

Factors involved in ignition of a solid

possible. For a flat solid heated in depth, the fuel mass flux leaving the surface could be described by Z py 00 As eEs =ðRTÞ dx ð7:2Þ m_ F ¼ 0

where py is a critical depth heated. This process is illustrated in Figure 7.1. We imply that before the surface temperature reaches Tpy there is no significant evolution of fuel 00 vapor. At Tpy , the value m_ F is sufficient to allow piloted ignition. Step 2 The evolved fuel vapor must be transported through the fluid boundary layer where the ambient air is mixed with the fuel. For piloted ignition, a sufficient energy source such as an electrical spark or small flame must be located where the mixed fuel is flammable, XF XL . For autoignition, we must not only achieve a near stoichiometric mixture but a sufficient volume of the vapor mixture must achieve its autoignition temperature (300–500 C). If the solid is radiatively heated, then either heat transfer from the surface or radiation absorption in the boundary layer must raise the temperature of the mixture to its autoignition value. This will be substantially higher than the typical surface temperature, Tpy , needed for piloted ignition. For a charring material, surface oxidation can occur before ignition in the gas phase results. This would be especially possible for the autoignition case or for low heat fluxes. For example, Spearpoint [1] saw thick wood ignite as low as 8 kW/m2 by radiant heat in the presence of a pilot. However, this flaming piloted ignition was preceded by surface glowing of the wood. Hence the energy of reaction for this oxidation augmented the radiant heating. Even without a pilot, glowing could induce flaming ignition. We shall ignore such glowing ignition and its effects in our continuing discussion, but it can be an important mechanism in the ignition of charring materials.

ESTIMATE OF IGNITION TIME COMPONENTS

161

Let us just consider the piloted ignition case. Then, at Tpy a sufficient fuel mass flux is released at the surface. Under typical fire conditions, the fuel vapor will diffuse by turbulent natural convection to meet incoming air within the boundary layer. This will take some increment of time to reach the pilot, whereby the surface temperature has continued to rise. Step 3 Once the flammable mixture is at the pilot, there will be another moment in time for the chemical reaction to proceed to ‘thermal runaway’ or a flaming condition. This three-step process can allow us to express the time for ignition as the sum of all three steps [2]: tig ¼ tpy þ tmix þ tchem

ð7:3Þ

where tpy ¼ conduction heating time for the solid to achieve Tpy tmix ¼ diffusion or transport time needed for the flammable fuel concentration and oxygen to reach the pilot tchem ¼ time needed for the flammable mixture to proceed to combustion once at the pilot

7.2 Estimate of Ignition Time Components The processes described by Equation (7.3) are complex and require an elaborate analysis to make precise determinations of tig . By making appropriate estimates of each of the time components, we can considerably simplify practical ignition analyses for most typical fire applications. Again, we are just considering piloted-flaming ignition but autoignition can be similarly described.

7.2.1

Chemical time

Consider the chemical time component, tchem . From Equation (5.22) we estimated the time for ignition of a flammable mixture for, say, a spherical volume of radius ro as tchem ¼ 0:21 ro2 =

ð7:4Þ

where is the thermal diffusivity of the gas mixture. For a runaway thermal reaction, the critical Damkohler number must be achieved for this sphere. Therefore, substituting for ro, E ro2 Ahc eE=ðRT1 Þ c ¼ ¼ 3:32 ð7:5Þ RT1 kT1 Combining, tchem ¼

ð0:21Þð3:3Þk T1 ½E=ðRT1 Þ A hc eE=ðRT1 Þ

ð7:6Þ

162

IGNITION OF SOLIDS

We estimate this time in the following manner (using Table 4.2), selecting E ¼ 160 kJ=mole A ¼ 1013 g=m3 s k ¼ 0:10 W=m K ¼ 16 106 m2 =s hc ¼ 45 kJ=g as typical flammable mixtures in air, and we select T1 ¼ 1600 Kð 1300 CÞ as the lowest adiabatic flame temperature for a fuel mixture in air to achieve flaming combustion. In general, we would have to investigate the energy release rate, Q_ chem ðtÞ ¼ A hc eE=ðRTÞ

4

3 3 ro

ð7:7Þ

and find t ¼ tchem when a perceptible amount of energy and light is observed at the pilot spark after the flammable mixture arrives. This is too cumbersome for illustrative purposes, so instead we select T1 ¼ 1600 K as a representative combustion temperature. Substituting these values, with E=ðRT1 Þ ¼ 12:0, tchem ¼

ð0:21Þð3:32Þð0:10 W=m KÞð1600 KÞ ð16 106 m2 =sÞð12:0Þð1013 g=m3 sÞð45 kJ=gÞe12:0

¼ 2:1 104 s: Thus, typical chemical reaction times are very fast. Note, if a chemical gas phase retardant is present or if the oxygen concentration is reduced in the ambient, A would be affected and reduced. Thus the chemical time could become longer, or combustion might not be possible at all.

7.2.2

Mixing time

Now let us consider the mixing time, tmix . This will be estimated by an order of magnitude estimate for diffusion to occur across the boundary layer thickness, BL . If we have turbulent natural conditions, it is common to represent the heat transfer in terms of the Nusselt number for a vertical plate of height, ‘, as hc ‘ ¼ 0:021ðGr PrÞ2=5 for Gr Pr > 109 k

ð7:8Þ

where Gr is the Grashof number, g½ðTs T1 Þ=T1 ð‘3 = 2 Þ for a gas, and Pr is the Prandtl number, =, with the kinematic viscosity of the gas. The heat transfer coefficient, hc , can be related to the gas conductivity k, approximately as hc

k BL

ð7:9Þ

ESTIMATE OF IGNITION TIME COMPONENTS

163

Then for a heated surface of ‘ ¼ 0:5 m, combining Equations (7.8) and (7.9), for typical air values at Ts ¼ 325 _ C and T1 ¼ 25 C ðT ¼ 175 CÞ Pr ¼ 0:69 ¼ 45 106 m2 =s ¼ 31 106 m2 =s k ¼ 35 103 W=m K 300 K ð0:5 mÞ3 ¼ 1:28 109 298 K ð31 106 m2 =sÞ2 BL 1 ¼ ¼ 0:0126 ‘ ð0:021Þ½ð1:28 109 Þð0:69Þ 2=5 Gr ¼ ð9:81 m=s2 Þ

or BL ¼ ð0:0126Þð0:5 mÞ ¼ 6:29 103 m The time for diffusion to occur across BL is estimated by pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ BL Dtmix

ð7:10Þ

where D is the diffusion coefficient, say 1:02 105 m2 =s for the laminar diffusion of ethanol in air. For turbulent diffusion, the effective D is larger; however, the laminar value will yield a conservatively, larger time. Equation (7.10) gives a diffusion penetration depth in time tmix . Therefore, we estimate tmix

ð6:29 103 mÞ2 ¼ 3:9 s 1:02 105 m2 =s

Hence, we see that this mixing time is also very small. However, this analysis implicity assumed that the pilot is located where the fuel vapor evolved. A pilot downstream would introduce an additional transport time in the process.

7.2.3

Pyrolysis

Finally, we estimate the order of magnitude of the time of the heated solid to achieve Tpy at the surface. This is primarily a problem in heat conduction provided the decomposition and gasification of the solid (or condensed phase) is negligible. We know that typically low fuel concentrations are required for piloted ignition ðXL 0:01–0:10Þ and by low 00 mass flux (m_ F 1–5 g=m2 s) accordingly. Thus, a pure conduction approximation is satisfactory. A thermal penetration depth for heat conduction can be estimated as T ¼

pﬃﬃﬃﬃﬃ t

ð7:11Þ

where is the thermal diffusivity of the condensed phase. We can estimate the pyrolysis time by determining a typical T .

164

IGNITION OF SOLIDS 00

The net incident heat flux at the surface is given by q_ . An energy conservation for the surface, ignoring any phase change or decomposition energies, gives @T 00 q_ ¼ k ð7:12Þ @x x¼0 At the pyrolysis time, we approximate the derivative as kðTpy T1 Þ @T k @x x¼0 T which allows an estimate of the pyrolysis time from Equation (7.11) as Tpy T1 2 tpy k c q_ 00

ð7:13Þ

ð7:14Þ

The property k c is the thermal inertia; the higher it is, the more difficult it is to raise the temperature of the solid. Let us estimate tpy for typical conditions for a wood product k c ¼ 0:5 ðkW=m2 KÞ2 s Tpy ¼ 325 C 00

q_ ¼ 20 kW=m2 ða relatively low value for ignition conditionsÞ Then from Equation (7.14), tpy

325 25 2 ð0:5Þ ¼ 113 s 20

Hence, the conduction time is the most significant of all, and it controls ignition. In general, thermal conductivity is directly related to density for solids, so k c 2 which emphasizes why lightweight solids, such as foamed plastics, can ignite so quickly. Also k and c usually increase as temperature increases in a solid. Hence the values selected for use in k c at elevated temperatures should be higher than generally given for normal ambient conditions.

7.3 Pure Conduction Model for Ignition Therefore, for piloted ignition we can say tig tpy and select a reasonable surface temperature for ignition. Although decomposition is going to occur, it mostly occurs at the end of the conduction time and is usually of short duration, especially for piloted ignition. The ignition temperature can be selected as the surface temperature corresponding to pure conductive heating as interpreted from experimental measurements. This is illustrated in Figure 7.2, and would apply to either

PURE CONDUCTION MODEL FOR IGNITION

Figure 7.2

165

Surface temperature response for ignition

a piloted or autoignition case. A higher Tig would result for autoignition with more distortion of a pure conductive heating curve near ignition. Nevertheless, a pure conduction model for the ignition of solids is a satisfactory approximation. Of course, information for Tig and thermal properties k; and c must be available. They can be regarded as ‘properties’ and could be derived as ‘fitting constants’ with respect to thermal conduction theory. In this regard, ignition temperature is not a fundamental thermodynamic property of the liquid flashpoint, but more likened to modeling parameters as eddy viscosity for turbulent flow, or for that matter kinematic viscosity ðÞ for a fluid approximating Newtonian flow behavior. The concept of using the ignition temperature is very powerful in implementing solutions, not only for ignition problems but for flame spread as well. Since the determination of Tig by a conduction model buries the processes of degradation and phase change under various possible heating conditions, we cannot expect a precisely measured Tig to be the same for all cases. Figure 7.3 shows results for the temperature of a wood particle board measured by K.

Figure 7.3 Piloted ignition of wood particle board (vertical) by radiant heating [3]

166

IGNITION OF SOLIDS

Saito with a fine wire thermocouple embedded at the surface [3]. The scatter in the results are most likely due to the decomposition variables and the accuracy of this difficult measurement. (Note that the surface temperature here is being measured with a thermocouple bead of finite size and having properties dissimilar to wood.) Likewise the properties k; and c cannot be expected to be equal to values found in the literature for generic common materials since temperature variations in the least will make them change. We expect k and c to increase with temperature, and c to effectively increase due to decomposition, phase change and the evaporation of absorbed water. While we are not modeling all of these effects, we can still use the effective properties of Tig ; k; and c to explain the ignition behavior. For example, tig increases as Tig increases (addition of retardants, more fire resistance polymers); tig increases as k; increase; tig increases as c increases (phase change, decomposition evaporation of water).

7.4 Heat Flux in Fire It would be useful to pause for a moment in our discussion of solids to put into perspective the heat fluxes encountered under fire conditions. It is these heat fluxes that promote ignition, flame spread and burning rate – the three components of fire growth.

7.4.1

Typical heat flux levels

Table 7.1 lists heat flux levels commonly encountered in fire and contrasts them with perceptible levels. It is typically found for common materials that the lowest heat fluxes to cause piloted ignition are about 10 kW/m2 for thin materials and 20 kW/m2 for thick materials. The time for ignition at these critical fluxes is theoretically infinite, but practically can be O(1 min) (order of magnitude of a minute). Flashover, or more precisely the onset to a fully involved compartment fire, is sometimes associated with a heat flux of 20 kW/m2 to the floor. This flow heat flux can be associated with typical Table 7.1

Common heat flux levels

Source

kW/m2

Irradiance of the sun on the Earth’s surface Minimum for pain to skin (relative short exposure)[4] Minimum for burn injury (relative short exposure)[4] Usually necessary to ignite thin items Usually necessary to ignite common furnishings Surface heating by a small laminar flame[5] Surface heating by a turbulent wall flame[6–8] ISO 9705 room–corner test burner to wall (0.17 m square propane burner) at 100 kW (0.17 m square propane burner) at 300 kW Within a fully involved room fire (800–1000 C) Within a large pool fire (800–1200 C)

1 1 4 10 20 50–70 20–40 40–60 60–80 75–150 75–267

167

HEAT FLUX IN FIRE

room smoke layer temperatures of 500 C (20 kW/m2 to ceiling) and 600 C (20 kW/m2 to the floor, as estimated by an effective smoke layer emissivity and configuration factor product of 0.6). The highest possible heat flux from the sun is not possible of igniting common solids. However, magnification of the sun’s rays through a fish bowl was found to have ignited thin draperies in an accidental fire.

7.4.2

Radiation properties of surfaces in fire

For radiant heating it is not sufficient to know just the magnitude of the incident heat flux to determine the temperature rise. We must also know the spectral characteristics of the source of radiation and the spectral properties (absorptivity, , transmittance, , and reflectance, ) of the material. Recall that þ þ ¼ 1

ð7:15Þ

and by equilibrium at a given surface temperature for a material, the emissivity is ¼

ð7:16Þ

Figures 7.4(a), (b) and (c) give the spectral characteristics of three sources: (a) the sun at the earth’s surface, (b) a JP-4 pool fire and (c) blackbody sources at typical room fire conditions (800–1100 K). The solar irradiance is mostly contained in about 0.3–2.4 m while fire conditions span about 1–10 mm. Figures 7.5 and 7.6 give the measured spectral reflectances and transmittances of fabrics. It is clear from Figure 7.5 that color (6,white; 7,black; 1,yellow) has a significant effect in reflecting solar irradiance, and also we see why these colors can be discriminated in the visible spectral region of 0.6 m. However, in the spectral range relevant to fire conditions, color has less of an effect. Also, the reflectance of dirty (5a) or wet (5b) fabrics drop to 0:1. Hence, for practical purposes in fire analyses, where no other information is available, it is reasonable to take the reflectance to be zero, or the absorptivity as equal to 1. This is allowable since only thin fabrics (Figure 7.6) show transmittance levels of 0.2 or less and decrease to near zero after 2 m.

7.4.3

Convective heating in fire

Convective heating in fire conditions is principally under natural convection conditions where for turbulent flow, a heat transfer coefficient of about 10 W/m2 K is typical. Therefore, under typical turbulent average flame temperatures of 800 C, we expect convective heat fluxes of about 8 kW/m2. Consequently, under turbulent conditions, radiative heat transfer becomes more important to fire growth. This is one reason why fire growth is not easy to predict.

168

IGNITION OF SOLIDS

Figure 7.4 Spectral characteristics of sources: (a) sun at Earth’s surface [9], (b) JP-4 fuel fire [10], (c) blackbodies [11]

HEAT FLUX IN FIRE

Figure 7.5

7.4.4

169

Spectral reflectance of cotton fabrics [11]

Flame radiation

Radiation from flames and combustion products involve complex processes, and its determination depends on knowing the temporal and spatial distributions of temperature, soot size distribution and concentration, and emitting and absorbing gas species concentrations. While, in principle, it is possible to compute radiative heat transfer if

Figure 7.6

Spectral reflectance and transmittance of fabrics [11]

170

IGNITION OF SOLIDS

these quantities are fully known, it has not become fully practical for the spectrum of realistic fire scenarios. Hence, only rational estimates, empirical correlations or measurements must be used. The difficulties can be appreciated by portraying the mean time heat flux emitted from a flame by g þ 0g ÞðT þ T 0 Þ4 q_ 00 ¼ ð

ð7:17Þ

where the overbars denote a time average and the primes the fluctuating turbulent components. Spatial variations, not considered here, would compound the complexity. In addition, the emissivity depends on fuel properties and flame shape. The flame or gas emissivity can be expressed in a simplified form as g ¼ 1 eg ‘

ð7:18Þ

where g is the absorption coefficient, which may typically be O(1) m1 for typical fuels in fire, and ‘ is a mean beam length or characteristic length scale for the fire [12]. Flames having length scales of 1–2 m would approach a blackbody emitter, g 1.

7.4.5

Heat flux measurements

Heat fluxes in fire conditions have commonly been measured by steady state (fast time response) devices: namely a Schmidt–Boelter heat flux meter or a Gordon heat flux meter. The former uses a thermopile over a thin film of known conductivity, with a controlled back-face temperature; the latter uses a suspended foil with a fixed edge temperature. The temperature difference between the center of the foil and its edge is directly proportional to an imposed uniform heat flux. Because the Gordon meter does not have a uniform temperature over its surface, convective heat flux may not be accurately measured. For a foil of radius, R, thickness, d, and conductivity, k, it can be shown that the Gordon gage temperature difference is

00 q_ r þ hc ðT1 TðRÞÞ Tð0Þ TðRÞ ¼ ½1 1=Io ðmRÞ hc 00

2 q_ r R þ ½T1 TðRÞ hc 4kd pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ 00 for small mR; m ¼ hc =kd ; Io is a Bessel function and q_ r and hc ðT1 TðRÞÞ are the net radiative and convective heat fluxes to the cooled meter at TðRÞ [6].

7.4.6

Heat flux boundary conditions

Heat flux is an important variable in fire growth and its determination is necessary for many problems. In general, it depends on scale (laminar or turbulent, beam length ‘), material (soot, combustion products) and flow features (geometric, natural or forced). We

IGNITION IN THERMALLY THIN SOLIDS

171

will make appropriate estimates for the heat flux as needed in quantitative discussions and problems, but its routine determination must await further scientific and engineering progress in fire. We will represent the general net fire heat flux condition as 00

q_ ¼

@T k @x x¼0 00

00

4 ¼ q_ e þ q_ f þ hc ðT1 Tð0; tÞÞ ðT 4 ð0; tÞ T1 Þ

ð7:19Þ

which gives the general surface boundary condition. With 00

q_ e ¼ external incident radiative heat flux 00 q_ f ¼ total (radiative þ convective) flame heat flux hc ðT1 Tð0; tÞÞ ¼ convective heating flux when no flame is present 4 ðT 4 ð0; tÞ T1 Þ ¼ net reradiative heat flux to a large surrounding at T1 Note that if this net flux is for a heat flux meter cooled at T1 , the ambient temperature, the gage directly measures the external incident radiative and flame heat fluxes.

7.5 Ignition in Thermally Thin Solids Let us return to our discussion of the prediction of ignition time by thermal conduction models. The problem reduces to the prediction of a heat conduction problem for which many have been analytically solved (e.g. see Reference [13]). Therefore, we will not dwell on these multitudinous solutions, especially since more can be generated by finite difference analysis using digital computers and available software. Instead, we will illustrate the basic theory to relatively simple problems to show the exact nature of their solution and its applicability to data.

7.5.1

Criterion for thermally thin

First, we will consider thin objects – more specifically, those that can be approximated as having no spatial, internal temperature gradients. This class of problem is called thermally thin. Its domain can be estimated from Equations (7.11) to (7.12), in which we say the physical thickness, d, must be less than the thermal penetration depth. This is illustrated in Figure 7.7. For the temperature gradient to be small over region d, we require d T

pﬃﬃﬃﬃﬃ kðTs To Þ t q_ 00

ð7:20aÞ

or Bi

dhc hc ðTs To Þ q_ 00 k

ð7:20bÞ

172

IGNITION OF SOLIDS

Figure 7.7

Thermally thin approximation

00

where Bi is the Biot number. If q_ is based just on convective heating alone, i.e. hc ðT1 Ts Þ; then Equation (7.20b) becomes Bi

Ts To T1 Ts

ðpure convectionÞ

ð7:21Þ

which says that for a thermally thin case – where d is small enough – the internal temperature difference must be much smaller than the difference across the boundary layer. Under typical ignition conditions for a solid fuel, k 0:2 W=m K Ts ¼ Tig 325 C To ¼ T1 ¼ 25 C 00

q_ ¼ 20 kW=m3 then d ð0:2 103 Þð325 25Þ=20 ¼ 3 mm Typically items with a thickness of less than about 1 mm can be treated as thermally thin. This constitutes single sheets of paper, fabrics, plastic films, etc. It does not apply to thin coatings or their laminates on noninsulating substances, as the conduction effects of the substrate could make the laminate act as a thick material. The paper covering on fiberglass insulation batting would be thin; pages in a closed book would not.

7.5.2

Thin theory

The theory of thermally thin ignition is straightforward and can apply to (a) a material of thickness d insulated on one side or (b) a material of thickness 2d heated symmetrically. The boundary conditions are given as @T 00 ðsurfaceÞ q_ ¼ k @x xs

173

IGNITION IN THERMALLY THIN SOLIDS

and

@T @x

¼0

ðcenter or insulated faceÞ

xi

However, by the thermally thin approximation, Tðx; tÞ TðtÞ only. A control volume surrounding the thin material with the conservation of energy applied, Equation (3.45), gives (for a solid at equal pressure with its surroundings)

cp d

dT 00 00 4 ¼ q_ ¼ q_ e hc ðT T1 Þ ðT 4 T1 Þ dt

ð7:22Þ

The net heat flux is taken here to represent radiative heating in an environment at T1 with an initial temperature T1 as well. From Equation (7.20) a more general form can apply if the flame heat flux is taken as constant. This nonlinear problem cannot yield an analytical solution. To circumvent this difficulty, the radiative loss term is approximated by a linearized relationship using an effective coefficient, hr : hr

4 ðT 4 T1 Þ 3 4T1 T T1

ð7:23Þ

for T T1 small. Hence, with ht hc þ hr , we have

cp d

dT 00 ¼ q_ e ht ðT T1 Þ dt

ð7:24Þ

Let ¼

T T1 q_ 00e =ht

ht t

cp d

ð7:25Þ

ð0Þ ¼ O

ð7:26Þ

and

Then d ¼1 d

with

which can be solved to give T T1

00 ht t q_ e ¼ 1 exp

cp d ht

(7.27)

00

for an insulated solid of thickness d heated by an incident radiative heat flux q_ e . In general, the flame heat flux can be included here. From Equation (7.27) as t ! 1, we reach a steady state, and as t ! 0; e 1 . This allows simple expressions for the ignition time, tig , which occurs when T ¼ Tig .

174

IGNITION OF SOLIDS

For small ignition times, since 1 e , tig

cp dðTig T1 Þ q_ 00e

ð7:28Þ

Note that the effect of cooling has been totally eliminated since we have implicitly 00 considered the high heating case where q_ e ht ðTig T1 Þ. Consider hc 0:010 kW=m2 K

4 hr 4ð1Þ 5:671 1011 kW=m2 K ð298 KÞ3 ¼ 0:006 kW=m2 K Tig ¼ 325 C Then, the heat flux due to surface cooling is ht ðTig T1 Þ ¼ ð0:010 þ 0:006Þð325 25Þ ¼ 4:8 kW=m2 00

Therefore this approximation might prove valid for q_ e > 5 kW/m2 at least. For long heating times, eventually at t ! 1, the temperature just reaches Tig . Thus for 00 any heat flux below this critical heat flux for ignition, q_ ig; crit , no ignition is possible by the conduction model. The critical flux is given by the steady state condition for Equation (7.27), 00

q_ ig; crit ¼ ht ðTig T1 Þ

ð7:29aÞ

or a more exact form (with ¼ 1), 00

4 Þ q_ ig; crit ¼ hc ðTig T1 Þ þ ðTig4 T1

7.5.3

ð7:29bÞ

Measurements for thin materials

Figures 7.8(a) and (b) display piloted ignition results for a metalized polyvinyl fluoride (MPVF) film of 0.2 mm thickness over a 25 mm fiberglass batting [14]. The MPVF film was bonded to a shear glass scrim (with no adhesive or significant thickness increase) to prevent it from stretching and ripping. The unbounded MPVF film was also tested. This shows several features that confirm the theory and also indicate issues. The ignition and rip-time data both yield critical heat fluxes of about 25 and 11 kW/m2 respectively. Clearly, both events possess critical heat fluxes that would indicate the

IGNITION IN THERMALLY THIN SOLIDS

175

Figure 7.8 (a) Piloted ignition of glass-bonded MPVF and ripping times for unbonded MPVF (d ¼ 0.2 mm) due to radiant heating and laid on fiberglass batting. (b) Reciprocal time results for ignition and ripping of thin MPVF (d ¼ 0:2 mm)

minimum needed to ignite or rip the film. These results allow the determination of Tig or Trip , critical temperatures, from Equation (7.29b). Only hc must be known for these determinations. 00 In Figure 7.8(b), the data are plotted as 1/t versus q_ e , and display a linear behavior as suggested by Equation (7.28). The linear results from the small time theory appear to hold for even ‘long times’; the measured times close to the critical heat fluxes are less than 60 s. Hence, this small time theory appears to hold over a practical range of ignition times. The applicability of Equation (7.28) over a wide time range is found to be generally true in practice. Moreover, the slope of these lines gives ½ cdðTig T1 Þ 1 , and allows a direct determination for c ( d is the face weight per unit area and can easily be measured). If this film (without bonding) is exposed to radiant heating, it would rip first. Depending on the final disposition of the ripped–stretched film, it may not ignite. Issues like this prevail in small-scale flame ignition tests, giving potentially misleading results. Thoughtful consideration of the fire scenario and its relevance to the test must be given.

176

IGNITION OF SOLIDS

7.6 Ignition of a Thermally Thick Solid The thermally thin case holds for d of about 1 mm. Let us examine when we might approximate the ignition of a solid by a semi-infinite medium. In other words, the backface boundary condition has a negligible effect on the solution. This case is termed thermally thick. To obtain an estimate of values of d that hold for this case we would want the ignition to occur before the thermal penetration depth, T reaches x ¼ d. Let us estimate this by d T

pﬃﬃﬃﬃﬃﬃﬃﬃ tig

ð7:30Þ

From experiments on common fuels in fire, the times for ignition usually range well below 5 minutes (300 s). Taking a representative ¼ 2 105 m2 =s, we calculate d

qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ ð2 105 m2 =sÞð300 sÞ ¼ 0:078 m

Hence, we might expect solids to behave as thermally thick during ignition and to be about 8 cm for tig ¼ 300 s and about 2.5 cm for 30 s. Therefore, a semi-infinite solution might have practical utility, and reduce the need for more tedious finite-thickness solutions. However, where thickness and other geometric effects are important, such solutions must be addressed for more accuracy.

7.6.1

Thick theory

Let us consider the semi-infinite (thermally thick) conduction problem for a constant temperature at the surface. The governing partial differential equation comes from the conservation of energy, and is described in standard heat transfer texts (e.g. Reference [13]): @T @2T ¼ 2 @t @x

ð7:31Þ

with boundary conditions: x ¼ 0;

T ¼ Ts ;

x ! 1;

constant

T ¼ T1

and the initial condition: t ¼ 0;

T ¼ T1

We will explore a solution of the form: T T1 ¼ ðÞ Ts T1

ð7:32Þ

IGNITION OF A THERMALLY THICK SOLID

177

where x ¼ pﬃﬃﬃﬃﬃ 2 t By substitution, it can be shown that @T d ¼ ðTs T1 Þ @t x d 2t 2 2 @ @T d 1 ¼ ðT T1 Þ 2 pﬃﬃﬃﬃﬃ @x @x d 2 t so that Equation (7.31) becomes d2 d þ 2 ¼ 0 d2 d

ð7:33Þ

with ¼ 0;

¼1

! 1;

¼0

This type of solution is called pﬃﬃﬃﬃﬃ a similarity solution where the affected domain of the problem is proportional to t (the growing penetration depth), and the dimensionless temperature profile is similar in time and identical in the variable. Fluid dynamic boundary layer problems have this same character. We solve Equation (7.33) as follows. Let ¼

d d

Then d ¼ 2 d Integrating, ln

¼ 2 þ c1

or ¼ c2 e

2

where we have changed the constant of integration. Integrating again gives ð0Þ ¼ c2

Z

2

e d 0

178

IGNITION OF SOLIDS

or ¼ 1 þ c2

Z

2

e d

0

Applying the other boundary condition, ! 1; ¼ 0, gives c2 ¼ R 1 0

1 e2 d

The integral is defined in terms of the error function given by 2 erfðÞ pﬃﬃﬃ

Z

e

2

d

ð7:34Þ

0

pﬃﬃﬃ where erfð1Þ ¼ 1 and erfð0Þ ¼ 0; c2 is found as ð2= Þ and the solution is ¼ 1 erfðÞ

ð7:35Þ

A more accurate estimate of T can now be found from this solution. We might define T arbitrarily to be x when ¼ 0:01, i.e. T ¼ T1 þ 0:01ðTs T1 Þ. This occurs where ¼ 1:82, or pﬃﬃﬃﬃﬃ T ¼ 3:6 t

ð7:36Þ

pﬃﬃﬃﬃﬃ If we reduce our criterion for T to x at ¼ 0:5, then T t, as we originally estimated. Table 7.2 gives tabulated values of the error function and related functions in the solution of other semi-infinite conduction problems. For example, the more general boundary condition analogous to that of Equation (7.27), including a surface heat loss,

@T k @x

00

00

¼ q_ ¼ q_ e ht ðT T1 Þ

ð7:37Þ

x¼0

(with the initial condition, t ¼ 0; T ¼ T1 ), gives the solution for the surface temperature ðTs Þ Ts T1 ¼ 1 expð 2 ÞerfcðÞ ðq_ 00e =ht Þ where ¼ ht

pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ t=ðk cÞ

(7.38)

IGNITION OF A THERMALLY THICK SOLID

179

Error function [13]

Table 7.2

erfðÞ

erfcðÞ

0 0.05 0.10 0.l5 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0

0 0.0564 0.112 0.168 0.223 0.276 0.329 0.379 0.428 0.475 0.520 0.563 0.604 0.642 0.678 0.711 0.742 0.771 0.797 0.821 0.843 0.880 0.910 0.934 0.952 0.966 0.976 0.984 0.989 0.993 0.995

1.00 0.944 0.888 0.832 0.777 0.724 0.671 0.621 0.572 0.525 0.489 0.437 0.396 0.378 0.322 0.289 0.258 0.229 0.203 0.170 0.157 0.120 0.090 0.066 0.048 0.034 0.024 0.016 0.011 0.007 0.005

2 p2ﬃﬃ e

1.128 1.126 1.117 1.103 1.084 1.060 1.031 0.998 0.962 0.922 0.879 0.834 0.787 0.740 0.591 0.643 0.595 0.548 0.502 0.458 0.415 0.337 0.267 0.208 0.159 0.119 0.087 0.063 0.044 0.030 0.021

The quantity erfc() is called the complementary error function, defined as erfcðÞ ¼ 1 erfðÞ

ð7:39Þ

Just as we examined the short ðt ! 0Þ and long ðt ! 1Þ behavior of the thermally thin solution, we examine Equation (7.38) for Ts ¼ Tig to find simpler results for tig. From the behavior of the functions by series expansions [13], ! large;

1 expð 2 Þ erfcðÞ pﬃﬃﬃ

ð7:40Þ

so that Equation (7.38) becomes Ts T1 expð 2 Þ expð 2 Þ 1 pﬃﬃﬃ ¼ 1 pﬃﬃﬃ 1 00 ðq_ e =ht Þ

ð7:41Þ

180

IGNITION OF SOLIDS

For the ignition time, at long times 00

k c ðq_ e =ht Þ2 tig ½q_ 00e ht ðTig T1 Þ 2

ð7:42Þ

As in the thermally thin case, as tig ! 1, the critical flux for ignition is 00

q_ ig;crit ¼ ht ðTig T1 Þ

ð7:43Þ

Similarly, investigating small-time behavior, we use the approximations from series expansions [13]: 2 small; erfðÞ pﬃﬃﬃ expð 2 Þ 1 þ 2 Then Ts T1 2 2 1 ð1 þ Þ 1 pﬃﬃﬃ q_ 00e =ht 2 2 ¼ þ pﬃﬃﬃ 2 þ pﬃﬃﬃ 3 2 pﬃﬃﬃ For the ignition time at small times, Tig T1 2 tig k c 4 q_ 00e

ð7:44Þ

Again as in the thermally thin case, the heat loss term is not included, suggesting that this 00 only holds where q_ e is large.

7.6.2

Measurements for thick materials

Figures 7.9(a) and (b) show piloted ignition data taken for a Douglas fir (96 mm square by 50 mm thick) irradiated by Spearpoint [1]. There is a distinction between samples heated across or along the grain (see Figure 7.10). Experimental critical ignition fluxes were estimated as 12:0 kW=m2 (across) and 9:0 kW=m2 (along); however, despite waiting for nearly 1 hour for ignition, a flux at which no ignition could be truly identified pﬃﬃﬃﬃﬃ was not determined. In Figure 7.9(b), the data are plotted as 1= tig against the incident radiation heat flux, as indicated by Equation (7.44). The ‘high’ heat flux data follow theory. The so-called ‘low’ heat flux data very likely contain surface oxidation, and are shifted to the left of the straight-line high heat flux data due to this extra heat addition. From the critical flux and the slope of the high heat flux line in Figure 7.9(b), Tig and k c

IGNITION OF A THERMALLY THICK SOLID

Figure 7.9

181

pﬃﬃﬃﬃﬃ 00 (a): Piloted ignition and (b) 1= tig versus q_ e for a 50 mm thick Douglas fir (from Spearpoint [1])

could be determined. These are given in Table 7.3 for several wood species tested by Spearpoint [1] along with their measured densities. In Figure 7.10 the anatomy of the virgin wood is depicted together with the definitions of heating ‘along’ and ‘across’ the grain used in Figure 7.9 and Table 7.3. For the ‘along the grain’ heating, it is easy for volatiles to escape the wood along the grain direction. This is likely to account for the lower Tig in the ‘along’ orientation. It is known that the conductivity parallel to the grain (along) is about twice that perpendicular to the grain (across). This accounts for the differences in the k c values.

182

IGNITION OF SOLIDS

Figure 7.10

7.6.3

Wood grain orientation

Autoignition and surface ignition

Boonmee [15,16] investigated the effect of surface oxidation observed by Spearpoint [1] at low heat flux for wood. Autoignition and the onset of surface ignition (‘glowing’) was studied for redwood under radiative heating. Glowing ignition was not based on a perception of color for the surface of the redwood, but by a distinct departure from inert heating. This process is essentially ignition to smoldering in contrast to flaming. At a high heat flux, the smoldering or wood degradation is controlled by oxygen diffusion, while at a low heat flux, it is controlled by decomposition chemical kinetics. The onset of flaming autoignition showed a distinct reliance on the energy supplied by smoldering at heat fluxes below about 40 kW/m2. These results are depicted in Figure 7.11 for the times for piloted ignition, autoignition and glowing ignition. There appears to be a critical flux for autoignition at 40 kW/m2, yet below this value flaming ignition can still occur after considerable time following glowing ignition. This glowing effect on autoignition is Table 7.3 Piloted ignition properties of wood species [1]a 00

Species

Heating

Density (kg/m3)

q_ ig;crit measured (kW/m2)

Redwood

Across Along Across Along Across Along Across Along

354 328 502 455 753 678 741 742

13.0 9.0 12.0 9.0 — — 12.0 8.0

Douglas fir Red oak Maple a

50 mm thick samples, moisture content 5–10%. Implies theoretically derived from data. 00 c Based on theoretical values of q_ ig . b

00

q_ ig;crit derivedb (kW/m2)

Tig ( C)c

15.0 6.0 16.0 8.0 11.0 9.0 14.0 4.0

375 204 384 258 305 275 354 150

k c

kW 2 s m2 K

0.22 2.1 0.25 1.4 1.0 1.9 0.67 11.0

IGNITION OF A THERMALLY THICK SOLID

Figure 7.11

183

Ignition times as a function of incident heat flux [15]

present to a measured critical flux of 20 kW/m2. The glowing transition to flaming cannot be predicted by a simple heat conduction model given by Equation (7.44). However, the remainder of the data follows conduction theory. Above 40 kW/m2, no perceptible differences in ignition times are found within the accuracy of the measurements. Ignition temperatures in the conduction theory regions of data suggest values indicated in Table 7.4, as taken for Spearpoint and Boonmee’s measurements given in Figure 7.12. Table 7.4 Estimated ignition temperatures and critical flux for redwood Heating along grain ( C) Piloted ignition Glowing ignition (<40 kW/m2) Autoignition (>40 kW/m2)

204 40080 35050

Heating across grain ( C) 375 48080 50050

Critical flux (kW/m2) 9–13 10 20

184

IGNITION OF SOLIDS

Figure 7.12

Flaming autoingition and glowing ignition temperatures as a function of incident heat flux [15]

Scatter in these data indicate accuracy, but they suggest that the earliest ignition time would be piloted, followed by glowing and then autoignition. This still may not preclude the effect of smoldering on piloted ignition at very low heat fluxes, as occurs with autoignition at 40 kW/m2. Table 7.4 clearly shows that the ignition temperature for autoignition is considerably higher than for piloted ignition. In conformance with this behavior, Boonmee [15] indicates that the corresponding fuel mass fractions needed are about 0.10 and 0.45 0.15 for the pilot and autoignition of redwood respectively.

7.7 Ignition Properties of Common Materials The values presented in Table 7.5 are typical of many common furnishings. A standard test method, ASTM E-1321 [17], has been developed to determine Tig and k c for materials and products. These items are treated as thermally thick in the analysis since they are backed by an insulating board in the test. They combine the thickness effect of the material tested with the thickness of the inert substrate in the standard test. Figure 7.13 shows examples of ignition and lateral flame spread rate data taken from materials as thin as 1.27 mm. The solid curve through the ignition data (to the right) follows the thermally thick theory (Equation (7.44)). The velocity of spread data plotted here are taken at long irradiance heating times, and are seen to be asymptotically symmetrical to the ignition time data about the critical flux for ignition. This implies that surface flame velocity is infinite when the surface is at the ignition temperature. In actuality, flame spread occurs here in the gas phase at the lower flammable limit.

185

IGNITION PROPERTIES OF COMMON MATERIALS Table 7.5

Piloted ignition and flame spread properties from ASTM 1321 [18] Tig ( C)

Material PMMA polycast, 1.59 mm Polyurethane (535M) Hardboard, 6.35 mm Carpet (acrylic) Fiberboard, low density (S119M) Fiber insulation board Hardboard, 3.175 mm Hardboard (S159M) PMMA Type G, 1.27 cm Asphalt shingle Douglas fir particle board, 1.27 cm Wood panel (S178M) Plywood, plain, 1.27 cm Chipboard (S118M) Plywood, plain, 0.635 cm Foam, flexible, 2.54 cm Glass/polyester, 2.24 mm Mineral wool, textile paper (S160M) Hardboard (gloss paint), 3.4 mm Hardboard (nitrocellulose paint) Glass/polyester, 1.14 mm Particle board, 1.27 cm stock Gypsum board, wall paper (S142M) Carpet (nylon/wool blend) Carpet #2 (wool, untreated) Foam, rigid, 2.54 cm Polyisocyanurate, 5.08 cm Fiberglass shingle Carpet #2 (wool, treated) Carpet #1 (wool, stock) Aircraft panel epoxy fiberite Gypsum board, FR, 1.27 cm Polycarbonate, 1.52 mm Gypsum board, common, 1.27 mm Plywood, FR, 1.27 cm Polystyrene, 5.08 cm

k c ((kW/m2K)2 s)

278 280 298 300 330 355 365 372 378 378 382 385 390 390 390 390 390 400 400 400 400 412 412 412 435 435 445 445 455 465 505 510 528 565 620 630

0.73 — 1.87 0.42 — 0.46 0.88 — 1.02 0.70 0.94 — 0.54 — 0.46 0.32 0.32 — 1.22 0.79 0.72 0.93 0.57 0.68 0.25 0.03 0.02 0.50 0.24 0.11 0.24 0.40 1.16 0.45 0.76 0.38

((kW)2/m3)

Ts;min ( C)

=ðk cÞ (m\K2/s)

5.45 — 4.51 9.92 — 2.25 10.97 — 14.43 5.38 12.75 — 12.91 — 7.49 11.70 9.97 — 3.58 9.81 4.21 4.27 0.79 11.12 7.32 4.09 4.94 9.08 0.98 1.83 — 9.25 14.74 14.44 — —

120 105 170 165 90 210 40 80 90 140 210 155 120 189 170 120 80 105 320 180 365 275 240 265 335 215 275 415 365 450 505 300 455 425 620 630

8 82 2 24 42 5 12 18 14 8 14 43 24 11 16 37 31 34 3 12 6 5 1 16 30 141 201 18 4 17 — 23 13 32 — —

Figure 7.14 shows the relationship of this standard test apparatus for Equation (7.29), 00 for which Ts ¼ functionðq_ e Þ: 00

Ts T1 00

00

00

4 ðTs4 T1 Þ q_ q_ ¼ e¼ e hc ht hc

ð7:45Þ

When q_ e ¼ q_ ig;crit , then Ts ¼ Tig , and hence the material properties can be determined. We see these ‘fitted’ effective properties listed in Table 7.5 for a variety of materials. Table 7.6 gives generic data for materials at normal room conditions. It can be seen that for generic plywood at normal room temperature k c ¼ 0:16ðkW=m2 KÞ2 s while under ignition conditions it is higher 0.5 (kW/m2 K)2 s. Increases in k and c with temperature can partly explain this difference. The data and property results were taken from Quintiere and Harkleroad [18].

186

IGNITION OF SOLIDS

Figure 7.13

Spread and ignition results for (a) plywood, (b) rigid polyurethane foam, (c) gypsum board and (d) asphalt shingle [18]

An alternative form of an ignition correlation used in ASTM E-1321 follows from Equations (7.44) and (7.45): 1=2 q_ 00ig;crit 4 h2t t ¼ ð7:46Þ ig 00 k c q_ e 00

where ht ¼ q_ ig;crit =ðTig T1 Þ. A result for fiberboard is shown in Figure 7.15.

187

IGNITION PROPERTIES OF COMMON MATERIALS

Figure 7.14

Equilibrium surface temperatures as a function of external radiant heating in the test apparatus [17]

Table 7.6

Thermal Properties of common materials at normal room temperature

Material

Density,

(kg/m3)

Concrete, stone Polymethyl methacrylate (PMMA) Gypsum board Calcium silicate board Particle board Plywood Cork board Balsa wood Polyurethane, rigid Polystyrene, expanded

2200 1200 950 700 650 540 200 160 32 20

a

Conductivity, k (W/m K) 1.7 0.26 0.17 0.11 0.11 0.12 0.040 0.050 0.020 0.034

Specific k c heat, c (kJ/kg K) ((kW/m2 K)2 s) 0.75 2.1 1.1 1.1 2.0 2.5 1.9 2.9 1.3 1.5

Compiled from various sources.

Figure 7.15

a

Correlation of ignition results for fiberboard [18]

2.8 0.66 0.18 0.085 0.14 0.16 0.015 0.023 0.0008 0.0010

188

IGNITION OF SOLIDS

References 1. Spearpoint, M.J., Predicting the ignition and burning rate of wood in the cone calorimeter using an integral model, MS Thesis, Department of Fire Protection Engineering, University of Maryland, College Park, Maryland 1999. 2. Fernandez-Pello, A.C., The solid phase, in Combustion Fundamentals of Fire, (ed. G. Cox), Academic Press, London, 1994. 3. Quintiere, J.G., The application of flame spread theory to predict material performance, J. Research of the National Bureau of Standards, 1988 93(1), 61. 4. Stoll, A.M. and Greene, L.C., Relationship between pain and tissue damage due to thermal radiation, J. Applied Physiology, 1959, 14, 373–82. 5. Ito, A. and Kashiwagi, T., Characterization of flame spread over PMMA using holographic interferometry, sample orientation effects, Combustion and Flame, 1988, 71, 189. 6. Quintiere, J.G., Harkleroad, M.F. and Hasemi, Y., Wall flames and implications for upward flame spread, J. Combus. Sci. Technol., 1986, 48(3/4), 191–222. 7. Ahmad, T. and Faeth, G.M., Turbulent wall fires, in 17th Symposium. (International) on Combustion, The Combustion Institute, Pittsburg, Pennsylvania, 1978, p. 1149. 8. Orloff, L., Modak, A.T. and Alpert, R.L., Turbulent wall fires, in 16th Symposium (International) on Combustion, The Combustion Institute, Pittsburg, Pennsylvania, 1979, pp. 1149–60. 9. Gast, P.R., Handbook of Geophysics, (ed. C.F. Campen), Macmillan Company, London, 1960, pp. 14–30. 10. Graves, K.W., Fire Fighter Exposure Study, Technical Report AGFSRS 71-1, Wright-Patterson Air Force Base, 1970. 11. Quintiere, J.G., Radiative characteristics of fire fighters’ coat fabrics, Fire Technol., 1974, 10, 153. 12. Tien, C.L., Lee, K.Y. and Stratton, A.J., Radiation heat transfer, The SFPE Handbook of Fire Protection Engineering, 2nd edn (eds P.J. DiNenno et al.), Section 1, National Fire Protection Association, Quincy, Massachusetts, 1995, p. 1-65. 13. Carslaw, H.S. and Jaeger, J.C., Conduction of Heat in Solids, 2nd edn, Oxford University Press, London, 1959, p. 485. 14. Quintiere, J.G., The effects of angular orientation on flame spread over thin materials, Fire Safety J., 2001, 36(3), 291–312. 15. Boonmee, N., Theoretical and experimental study of the auto-ignition of wood, PhD Dissertation, Department of Fire Protection Engineering, Univerity of Maryland, College Park, Maryland, 2004. 16. Boonmee, N. and Quintiere, J.G., Glowing and flaming auto-ignition of wood, in 29th Symposium (International) on Combustion, The Combustion Institute, Pittsburg, Pennsylvania, 2002, pp. 289–96. 17. ASTM E-1321, Standard Test Method for Determining Material Ignition and Flame Spread Properties, American Society for Testing and Materials Philadelphia, Pennsylvania, 1996. 18. Quintiere, J.G. and Harkleroad, M.F., New concepts for measuring flame spread properties, in Fire Safety Science and Engineering, ASTM STP882, (ed. T.-Z. Harmathy), American Society for Testing and Materials, Philadelphia, Pennsylvania, 1985, p. 239.

Problems 7.1

The time for ignition of solid fuels is principally controlled by

(a) The heating time of the solid (b) The time for diffusion in air (c) The reaction time in the gas

Yes

Not necessarily

No

____ ____ ____

___ ___ ___

___ ___ ___

189

PROBLEMS (d) Ignition temperature (e) Density

____ ____

___ ___

7.2

Explain how the time to ignition of a solid can be based on conduction theory.

7.3

Kerosene as a thick, motionless fluid is in a large tank.

___ ___

The overall heat transfer coefficient for convective and radiative cooling at the surface is 12 W/m2 K. The thermal conductivity of the kerosene is 0.14 W/m K. Density of kerosene is 800 kg/m3. Specific heat of liquid kerosene is 1.2 J/g K. Other properties are given in Table 6.1 as needed. The ambient temperature is 25 C. If the incident heat flux is 10 kW/m2, how long will it take the kerosene to ignite with a pilot flame present? 7.4

A flame impinges on a ceiling having a gas temperature of 800 C and a radiation heat component of 2 W/cm2 to the ceiling. The ceiling is oak. How long will it take to ignite? Assume h ¼ 30 W/m K, T1¼ 20 C and use Tables 7.4 and 7.5.

7.5

Calculate the time to ignite (piloted) for the materials listed below if the irradiance is 30 kW/ m2 and the initial temperature is 25 C. The materials are thick and the convective heat transfer coefficient is 15 W/m2 K. Compute the critical flux for ignition as well. Material PMMA Fiberboard Plywood Wood carpet Polyurethane foam Paper on gypsum board

7.6

Tig ðdeg CÞ 380 330 390 435 435 565

k cððkW=m2 KÞ2 Þs 1.0 0.46 0.54 0.25 0.03 0.45

A thin layer of Masonite (wood veneer) is attached to an insulating layer of glass wool. The Masonite is 2 mm thick. It has the following properties: k ¼ 0:14 W=m K

¼ 0:640 g=m3 cp ¼ 2:85 J=gK Tig ¼ 300 C The total radiative–convective heat transfer coefficient in still air at 25 C is 30 W/m2 K. (a) Calculate the time to ignition for Masonite when subjected to a radiant heat flux of 50 kW/m2. (b) What is the critical or minimum heat flux for ignition? (c) After ignition, what is the initial upward spread rate if the flame heat flux is uniform at 30 kW/m2 and the flame extends 0.2 m beyond the ignited region of 0.1 m?

7.7

A flame radiates 40 % of its energy. The fuel supply is 100 g/s and its heat of combustion is 30 kJ/g. A thin drapery is 3 m from the flame. Assume piloted ignition. When will the drapery ignite? The ambient temperature is 20 C and the heat transfer coefficient of the drapery is

190

IGNITION OF SOLIDS 10 W/m2 K. The drapery properties are listed below:

¼ 40 kg=m3 cp ¼ 1:4 J=g K Tig ¼ 350 C Thickness ¼ 2mm ðassume thinÞ The incident flame heat flux can be estimated as Xr Q_ =ð4r 2 Þ, where Xr is the flame radiation fraction and r is the distance from the flame.

7.8

A vertical block of wood is heated to produce volatiles. The volatiles mix with entrained air in the boundary layer as shown below. A spark of sufficient energy is placed at the top edge of the wood within the boundary layer. The temperature can be assumed uniform across the boundary layer at the spark. The entrained air velocity (va), air density ( a) and temperature (Ta) are given on the figure. All gases have fixed specific heat, cp¼1 J/g K. The wood is 10 cm wide by 10 cm high. The surface temperature (Ts) of the wood reaches 400 C.

Using the assumption of a minimum flame temperature needed for ignition of the mixture, determine the minimum fuel mass loss rate per unit surface area ðm_ 00f Þ to cause flame propagation through the boundary layer. The heat of combustion that the volatile wood produces ðhc Þ is 15 kJ/g. (Hint: the adiabatic flame temperature at the lower flammable limit for the mixture in the boundary layer must be at least 1300 C.) 7.9

The temperature at the corner edge of a thick wood element exposed to a uniform heat flux can be expressed, in terms of time, by the approximate equation given below for small time: 4:2 00 pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ Ts T1 pﬃﬃﬃ q_ e t=k c The notation is consistent with that of the text. The properties of the wood are: Thermal conductivity Density Specific heat Piloted ignition temperature Autoignition temperature Ambient temperature

0.12 W/m K 510 kg/m3 1380 J/kg K 385 C 550 C 25 C

A radiant heat flux of 45 kW/m2, alone, is applied to the edge of the wood. When will it ignite? 7.10

A thin piece of paper is placed in an oven and raised to 60 C. When taken out of the oven, it is subjected to 30 kW/m2 and ignites in 50 s. It is known to ignite in 130 s, when at 20 C and similarly subjected to 30 kW/m2. What is its ignition temperature?

8 Fire Spread on Surfaces and Through Solid Media

8.1 Introduction In Chapter 4 we examined the spread rate of a premixed flame and found that its speed, 000 Su , depended on the rate of chemical energy release, m_ F hc . Indeed, for a laminar flame, the idealized flame speed 000

Su

m_ F hc R cp ðTf T1 Þ

ð8:1Þ

where Tf ¼ flame temperature T1 ¼ initial temperature of the mixture R ¼ thickness of the chemical reaction zone In words we can write Flame speed ¼

½chemical energy release rate ½energy needed to raise the substance to its ignition temperature

ð8:2Þ

We substitute ‘ignition temperature’ for ‘flame temperature ðTf Þ’ because we are interested in the speed of the pyrolysis (or evaporation) front ðvp Þ just as it ignites. As we have seen from Chapter 7 on the ignition of solids, this critical surface temperature can be taken as the ignition temperature corresponding to the lower flammable limit. Indeed, we might use the same strategy of eliminating the complex chemistry of decomposition and phase changes to select an appropriate ‘ignition temperature’. If we use a conduction-based model as in estimating the ignition of solids, then the same measurement techniques for identifying Tig can prevail.

Fundamentals of Fire Phenomena James G. Quintiere # 2006 John Wiley & Sons, Ltd ISBN: 0-470-09113-4

192

FIRE SPREAD ON SURFACES AND THROUGH SOLID MEDIA

Although very fundamental theoretical and experimental studies have been done to examine the mechanisms of spread on surfaces (solid and liquids), a simple thermal model can be used to describe many fire spread phenomena. This was eloquently expressed in a paper by Forman Williams [1] in discussing flame spread on solids, liquids and smoldering through porous arrays. Excellent discussions on flame spread and fire growth can be found in reviews by Fernandez-Pello [2] and by Thomas [3]. Some noteworthy basic studies of surface flame spread against the wind include the first extensive solution by deRis [4], experimental correlations to show the effect of wind speed and ambient oxygen concentration by Fernandez-Pello, Ray and Glassman [5,6] and the effect of an induced gravitation field by Altenkirch, Eichhorn and Shang [7]. For surface flame spread in the same direction as the wind (or buoyant flow), laminar cases have been examined by Loh and Fernandez-Pello [8,9] with respect to forced flow speed and oxygen concentration, and by Sibulkin and Kim [10] for natural convection laminar upward flame spread. Many features have been investigated for flame spread. These include orientation, wind speed, gravitational variations, laminar and turbulent flows, solid, liquid or porous medium, flame size and radiation, and geometrical effects – corners, channels, ducts and forest terrain to urban terrains. Figure 8.1 illustrates several modes of flame spread on a solid surface. We can have wind-aided (or concurrent) flame spread in which the pyrolysis front at xp moves in the directions of the induced buoyant or pressure-driven ambient flow speed, u1 . Consistent with the thermal model, we require, at x ¼ xp , Ts to be equal to the ignition temperature, Tig . In general, Ts ðxÞ is the upstream or downstream surface temperature of the virgin material. A wind-directed flame on the floor will be different than one on the ceiling. In the floor case, the larger the fire becomes, the more its buoyant force grows and tries to make the flame become vertical despite the wind. This phenomenon will be profoundly different in a channel than on an open floor. In the ceiling case, the gravity vector helps to hold the flame to the ceiling, and even tries to suppress some of its turbulence.

Figure 8.1

Modes of surface flame spread

INTRODUCTION

Figure 8.2

193

Dynamics of opposed flow surface flame spread

When the ambient flow is directed into the advancing flame, we call this case opposed flow flame spread. The flow can be both natural and artificially forced. We will see that opposed flow spread is much slower than wind-aided spread and tends to be steady. This can easily be seen by observing downward or lateral (horizontal) flame spread on a matchstick. However, invert the match to allow the flame to spread upward and we have an almost immeasurable response (provided we can hold the match!). In opposed flow spread, we know that the flame will be quenched near the surface since the typical range of ignition temperatures (250–450 C) is not likely to support gas phase combustion. Hence, this gap will allow air and fuel to meet before burning, so a premixed flame is likely to lead the advancing diffusion flame, as illustrated in Figure 8.2. If the ambient speed u1 ¼ 0, it will still appear to an observer on the moving flame (or the pyrolysis leading edge position, xp ) that an opposed flow is coming into the flame at the flame speed, vp . Even under zero gravity conditions, this is true; therefore, opposed flow spread may play a significant role in spacecraft (microgravity) fires. As u1 increases, mixing of 000 fuel and oxygen can be enhanced and we expect a higher m_ F and therefore we expect vp to increase. However, at high speeds the flow time of air or oxygen through the reaction zone ðR Þ may give a flow time tflow ¼ R =u1 < tchem . Thus, the speed of oxygen through the leading premixed flame may be so great as to inhibit the chemical reaction rate. This can lead to a decrease in vp . Similar arguments can describe the effect of ambient oxygen 000 concentration on vp ; i.e. as YO2 ;1 increases, m_ F increases and so vp increases. We shall examine some of these effects later in a more explicit thermal model for vp. Since accidental fire spread mostly occurs under natural convection conditions within buildings and enclosures, some examples of configurations leading to opposed or windaided types of spread are illustrated in Figure 8.3. Flame spread calculations are difficult

Figure 8.3

Examples of surface flame spread under natural convective conditions

194

FIRE SPREAD ON SURFACES AND THROUGH SOLID MEDIA

because of the various flow conditions that can occur. Moreover, the size and scale of the fire introduce complexities of turbulence and radiation heat transfer. We will not be able to deal with all of these complexities directly. There are no practical and available solutions that exist for every case. In this chapter we will present flame spread theory in its simplest form. However, we will illustrate how practical engineering estimates can be made for materials based on data and empirical relationships. While we will not address forest fire spread, a similar practical engineering approach has proven useful in the defense of forest fires. It relies on forest vegetation data, humidity, wind, slope of terrain and empirical formulas [11]. We will not describe all these forest fire details, but we will address some aspects of flame spread through porous material, such as fields of grass, forest litter, etc. In this chapter we will also touch on flame spread over liquids and fire growth in multiroom dwellings. The scope of fire spread is very wide; this is why the issue of ‘material flammability’ is so complex.

8.2 Surface Flame Spread – The Thermally Thin Case We will develop an analytical formulation of the statement in Equation (8.2). This will be done for surface flame spread on solids, but it can be used more generally [1]. As with the ignition of solids, it will be useful to consider the limiting cases of thermally thin and thermally thick solids. In practice, these solutions will be adequate for first-order approximations. However, the model will not consider any effects due to (a) melting: involving dripping, heats of melting and vaporization; (b) charring: causing an increase in surface temperature between pyrolysis and ignition; (c) deformation: manifested by stretching, shrinking, ripping, curling, etc.; (d) inhomogeneity: involving the effects of substrate, composites, porosity and transparency absorption to radiation. Factors (a), (b) and (d) are all swept into the property parameters: Tig and k, and c. Deformation (c) is ignored, but may be significant in practice. A curling material could enhance spread, while ripping could retard it. Dripping in factor (a) could act in both ways, depending on the orientation of the sample and direction of spread. Any test results that lead to the thermal properties for ignition and those to be identified for spread must apply to the fire scenario for the material in its end use state and configuration. For example, a test of wallpaper by itself as a thin sheet would not suffice for evaluating its flame spread potential when adhered to a gypsum wall board in a building corridor subject to a fully involved room fire. Figure 8.4 displays a thermally thin solid of thickness d, insulated at its back face and undergoing surface flame spread. Any of the modes shown in Figure 8.1 are applicable. We consider the case of steady flame spread, but this constraint can be relaxed. The surface flame spread speed is defined as vp ¼

dxp dt

ð8:3Þ

SURFACE FLAME SPREAD – THE THERMALLY THIN CASE

195

Figure 8.4 Thermal model for surface flame spread

In Figure 8.4 the control volume surrounds the solid and is fixed at x ¼ xp and a distance far enough ahead of the flame heat transfer affected region of the solid. The possible heat fluxes include: 00

q_ f flame incident convective and radiative heat flux 00 q_ k;p conduction from the pyrolyzed region 00 q_ k;1 , conduction from the CV to Ts Here Ts is taken to mean the surface temperature of the solid that is not affected by the 00 flame heating. If Ts is constant or varies slowly with xð@T=@x 0Þ, we can set q_ k;1 ¼ 0. 00 Much study has gone into assessing the importance of q_ k;p , and it can be significant. 00 However, the dominant heat transfer, even under slow opposed spread, is q_ f . For concurrent spread this approximation is very accurate since the flame surface heating distance is very long and @T=@x 0 near x ¼ xp. We shall adopt the solid temperature necessary for piloted ignition as the ignition temperature. This is in keeping with the initiation of pyrolysis to cause ignition of the evolved fuel with a pilot being the advancing flame itself. We apply Equation (3.45) for the CV fixed to x ¼ xp . The solid with density and specific heat cp appears to enter and leave the CV with velocity vp, the flame speed. Thus, cp dvp ðTig Ts Þ ¼

Z

1h

i 00 4 Þ dx q_ f ðxÞ ðT 4 ðxÞ T1

xp

ð8:4Þ

where we have approximated the surface as a blackbody and consider radiation to an 00 4 ambient at T1 . Since we expect q_ f ðTig4 T1 Þ we will ignore this nonlinear 00 temperature term. Furthermore, we define an average or characteristic q_ f for an effective heating length f as 00

q_ f f

Z

1

xp

00

q_ f ðxÞ dx

ð8:5Þ

Physically f is the flame extension over the new material to ignite. This is illustrated for a match flame, spreading upward or downward in Figure 8.5 or as suggested in Figure 8.1. We shall see that the visible flame is a measure for f because at the flame tip in a turbulent flame the temperature can drop from 800 C in the continuous luminous zone to

196

FIRE SPREAD ON SURFACES AND THROUGH SOLID MEDIA

Figure 8.5

Flame heating extensions for a match flame

about 350 C. Hence, a point between the continuous luminous zone and the fluctuating flame tip might constitute an approximation for f . In any case, Equation (8.5) is 00 mathematically acceptable for a continuous function, q_ f ðxÞ, according to the mean value theorem. Equations (8.4) and (8.5) yield a solution 00

vp ¼

q_ f f cp dðTig Ts Þ

ð8:6Þ

This can be put into a different form to reveal the relationship of flame spread to ignition: f tf cp dðTig Ts Þ tf ¼ 00 q_ f

vp ¼

(8.7a) (8.7b)

This is identical with our result for the ignition time at a high heat flux in the thermally thin ignition of a solid. It can be generalized to say that flame speed – thermally driven – can be represented as the ratio of the flame extension length to the time needed to ignite this heated material, originally at Ts . Equation (8.6) demonstrates that as the face weight, d, decreases the spread rate increases. Moreover, if a material undergoing spread is heated far away from the flame, such as would happen from smoke radiation in a room fire, Ts will increase over time. As Ts ! Tig , an asymptotic infinite speed is suggested. This cannot physically happen. Instead, the surface temperature will reach a pyrolysis temperature sufficient to cause fuel vapor at the lower flammable concentration. Then the speed of the visible flame along the surface will equal the premixed speed. This speed in the gas phase starts at about 0.5 m/s

SURFACE FLAME SPREAD – THE THERMALLY THIN CASE

Figure 8.6

197

Enhanced flame spread due to preheating

and can accelerate due to turbulence (see Figure 8.6). We see under these conditions why smoke-induced radiant heat flux of about 20 kW/m2 might provoke rapid flame spread across objects on a floor during a room fire. This mechanism can result in flashover – a rapid transition to a fully involved room fire. It should be pointed out that Equation (8.6), and its counterpart for thermally thick materials, will hold only for Ts > Ts;min, a minimum surface temperature for spread. Even if we include the heat loss term in Equation (8.4) by a mean-value approximation for the integrand, Tig4 þ Ts4 2 we obtain a physically valid solution still possible for all Ts as 00 q_ f f ðTig4 þ Ts4 Þ=2 f vp ¼ cp dðTig Ts Þ

ð8:8Þ

as long as the numerator is positive. Usually at Ts ¼ Ts;min , vp > 0, as determined in the experiments. This is shown in Figures 8.7 and 8.8 for lateral spread and upward spread on a vertical surface (for thermally thick behavior of PMMA and wood respectively). The

Figure 8.7

Natural steady flame speed and piloted ignition for PMMA (vertical) (from Long [12])

198

Figure 8.8

FIRE SPREAD ON SURFACES AND THROUGH SOLID MEDIA

Pyrolysis-front propagation velocity vp, and flame-tip propagation velocity vf, as functions of the incident radiant energy flux, for a preheat time tp of 2 min on a wood sample [13]

incident external radiant heat flux is a surrogate here for the steady state Ts at equilibrium. Thus, there is a critical heat flux for flame spread (as in ignition) given by 00

4 4 T1 Þ þ hc ðTs;min T1 Þ q_ s;crit ¼ ðTs;min

ð8:9Þ 00

This follows by a steady state energy balance of the surface heated by q_ e, outside the flame-heated region f . It appears that a critical temperature exists for flame spread in both wind-aided and opposed flow modes for thin and thick materials. Ts;min has not been shown to be a unique material property, but it appears to be constant for a given spread mode at least. Transient and chemical effects appear to be the cause of this flame spread limit exhibited by Ts;min. For example, at a slow enough speed, vp , the time for the pyrolysis may be slower than the effective burning time: tf ¼

f > tb vp

ð8:10Þ

and the inequality in this directly leads to extinction. For wind-aided spread the limit condition will depend on the flame extension and on the burning time; for opposed flow spread, the ability to achieve a flammable mixture at the flame front is key. In addition, transient effects will influence all of these factors.

8.3 Transient Effects It is useful to consider the transient aspect of the flame spread problem. The ratio of characteristic times for the solid and the gas phase, based on d as a characteristic length scale, is tgas d 2 = gas

solid 107 m2 =s ¼ 2 ¼ 5 2 ¼ 102 10 m =s tsolid d = solid

gas

TRANSIENT EFFECTS

199

as a typical value. Hence, the gas phase response is very fast and can adjust quickly to 00 any transient changes in solid. Thus, any steady state heat transfer data for q_ f or flame extension correlations (e.g. for upward, ceiling or wind-aided flames) can be directly applied to the transient problem of flame spread on the solid. However, we have only considered a steady state solid model in Equation (8.4). Consider the effect if we add the transient term (Equation (3.40)), the rate of enthalpy per unit area in the CV, Z dH 00 d x¼xf ðtÞ dcp Tdx ð8:11Þ dt x¼xp ðtÞ dt where xf is the position of the flame tip, xp þ f . Here T1 is a reference temperature, say 25 C. For opposed flow spread we do not expect f to depend on xp , the region undergoing pyrolysis. Here we consider that x ¼ 0 to x ¼ xp is pyrolyzing, although burnout may have occurred for x ¼ xb < xp. We will ignore the burnout effect here. For opposed flow spread, f will depend on the fluid mechanics of the induced flow ðu1 Þ in combination with the slowly opposite moving flame ðvp Þ. However, for wind-aided spread it might be expected that f will depend on the extent of the pyrolysis region – in other words, a longer flame for more fuel produced over x ¼ 0 to x ¼ xp . The transient term can be approximated as follows, where cp cv : Tig þ Ts dH 00 d dcp f dt dt 2 Tig þ Ts df dTs f ¼ dcp þ ð8:12Þ dt 2 2 dt If we consider cases where Ts is changing slowing with time, then dTs 0 dt is reasonable. For opposed flow, f 6¼ f ðxp ðtÞÞ, so df =dt ¼ 0. However, for the example 00 of wind-aided (upward) spread in Figure 8.8 it appears that df =dt > 0 as q_ e increases, since df ¼ vf vp dt Thus, for opposed flow spread, the steady state thermal flame spread model appears valid. In wind-aided flame spread, it seems appropriate to modify our governing equation for the thermally thin case as Tig þ Ts df 00 dcp þ dcv vp ðTig Ts Þ ¼ q_ f f 2 dt

ð8:13Þ

df dxf dxp dxf ¼ ¼ vp 1 dt dt dt dxp

ð8:14Þ

Since

200

FIRE SPREAD ON SURFACES AND THROUGH SOLID MEDIA

combining gives, for wind-aided spread on a thin material, 00

vp

q_ f f i Tig þTs dxf dcp ðTig Ts Þ þ dx 1 2 p h

ð8:15Þ

From Figure 8.8, vf > vp , so dxf =dxp > 1. Hence, transient effects will cause a lower speed for wind-aided spread. Again, since the gas phase response time is much faster than 00 the solid, we can use steady gas phase results for q_ f and f in these formulas for flame spread on surfaces.

8.4 Surface Flame Spread for a Thermally Thick Solid Let us now turn to the case of a thermally thick solid. Of course thickness effects can be important, but only after the thermal penetration depth due to the flame heating reaches the back face, i.e. t ¼ tf , T ðtf Þ ¼ d. As in the ignition case, if tf is relatively small, say 10 to even 100 s, the thermally thick approximation could even apply to solids of d 1 cm. Again, we represent all of the processes by a thermal approximation involving the effective properties of Tig , k, and c. Materials are considered homogeneous and any measurements of their properties should be done under consistent conditions of their use. Other assumptions for this derivation are listed below: 00

1.

q_ f is constant over region f and zero beyond.

2.

Negligible surface heat loss, i.e. q_ f Tig4 .

3.

At y ¼ T , q_ ¼ 0 and T ¼ T1 .

4.

Both initial and upstream temperatures are T1 .

5.

Steady flame spread, vp ¼ constant.

00

00

Consider the control volume in Figure 8.9 where the thermally thick case is drawn for a wind-aided mode, but results will apply in general for the opposed case. The control

Figure 8.9

Control volume energy conservation for a thermally thick solid with flame spread steady velocity, VP

SURFACE FLAME SPREAD FOR A THERMALLY THICK SOLID

201

volume is selected to span the flame heated length f and extend to the depth T . It is fixed to the point, P. The specific depth is T ¼ T ðtf Þ, since tf is the time for the temperature at P to have been increased from T1 to Tig . However, tf is also the time for the flame to traverse f , i.e. Z tf vp dt ¼ vp tf ð8:16Þ f ¼ 0

since vp is constant. The conservation of energy for the CV is Z cp vp 0

T

00

ðT T1 Þ dy ¼ q_ f f

ð8:17Þ

Let us approximate the temperature profile as T T1 ¼ Tig T1

y 2 1 T

ð8:18Þ

which satisfies y ¼ 0;

T ¼ Tig

y ¼ T ;

T ¼ T1 and

@T ¼0 @y

This approximation gives us a reasonable profile that satisfies the boundary conditions and allows us to perform the integration Z

T

ðT T1 Þ dy ¼ ðTig T1 Þ T =3

ð8:19Þ

0

We have seen that a reasonable approximation for the penetration depth is sﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ k T ¼ C t cp

ð8:20Þ

where C can range from 1 to 4 depending on its definition of how close T is to T1 . Let us take C ¼ 2:7 as a rational estimate. Now we substitute the results of Equations (8.16), (8.19) and (8.20) into Equation (8.17) to obtain ðTig T1 Þ 2:7 cp vp 3

sﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ k f 00 q_ f f cp vp

or 00

vp

ðq_ f Þ2 f ð0:81Þðkcp ÞðTig T1 Þ2

ð8:21Þ

202

FIRE SPREAD ON SURFACES AND THROUGH SOLID MEDIA

Returning to the form of Equation (8.16), we can estimate Tig T1 2 tf 0:81 kcp 00 q_ f

ð8:22Þ

We might just as well use the exact thermally thick high heat flux (short time) result, where 0.81 is replaced by =4 ¼ 0:785. We can use 00

vp

4 ðq_ f Þ2 f ðkcp ÞðTig Ts Þ2

ð8:23Þ

where we have generalized to Ts which could be changing under smoke and fire 00 conditions. Thus Equation (8.23) might be used in a quasi-steady form where q_ f ; f , and Ts might all vary with time as the flame grows and fire conditions worsen. Furthermore, Equation (8.23) follows the form of Equation (8.7) which says that flame spread is essentially the flame heating distance divided by the ignition time.

8.5 Experimental Considerations for Solid Surface Spread We can adopt Equations (8.6) and (8.23) for practical applications. Since these results complement the corresponding thermally thin and thick ignition solutions, the thermal properties should be identical for both ignition and flame spread with respect to a given material.

8.5.1

Opposed flow

For opposed flow spread under natural convection conditions, Ito and Kashiwagi [14] 00 have measured the q_ f ðxÞ profile for PMMA (d ¼ 0:47 cm) at several angles. Their results, shown in Figure 8.10, suggest values for 00

q_ f 70 kW=m2 and f 2 mm as rough approximations for their experimental profiles, e.g. ¼ 90 . Furthermore, f increases as spread goes from downward ð ¼ 0 Þ to horizontal ð ¼ 90 Þ to slightly upward ð ¼ 80 Þ. These angles all constitute natural convection opposed flow flame spread. A more practical way to yield results for opposed flow spread is to recognize that the parameter

4 00 2 ðq_ Þ f f

ð8:24Þ

EXPERIMENTAL CONSIDERATIONS FOR SOLID SURFACE SPREAD

Figure 8.10

203

Distributions of normal heat flux at the surface along the distance ahead of the pyrolysis front for 0.47 cm thick PMMA [14]

can be regarded as a material property for a given flow condition. This can be argued from laminar premixed conditions for the leading flame with the heat flux related to flame temperature and distances related to a quenching length. Under natural convection induced flow velocities, might be constant. Figure 8.11 shows steady velocity data plotted for thick PMMA as a function of incident radiant heat flux [12,15].

Figure 8.11

Evaluation of from the LIFT [12]

204

FIRE SPREAD ON SURFACES AND THROUGH SOLID MEDIA

Under steady conditions,

vp ¼

; kcp ðTig Ts Þ2

ð8:25Þ

Ts > Tmin

If Ts is caused by long-time radiant heating, we can write 00

q_ e ¼ ht ðTs T1 Þ

ð8:26Þ

For ht approximated as a constant, we can alternatively write Equation (8.25) as v1=2 ¼ p

kcp 1=2 00 00 ðq_ ig;crit q_ e Þ 2 ht

ð8:27Þ

00

where q_ ig;crit is the critical heat flux for ignition given by Equation (7.28). Since kcp can 00 be evaluated from ignition data for the material, and for ht evaluated at q_ ig;crit , then can be determined from the slope of the data in Figure 8.11. Also the intercept in the abscissa 00 00 is q_ ig;crit and the lower flux limit gives q_ s;crit or Ts;min from Figure 8.11. Data for a number of materials under conditions of lateral flame spread for a vertical sample in the LIFT [15,16] are given in Table 8.1 (also see Table 7.4). The value of given for PMMA is 14:0 kW2 =m3 , which can be compared to an estimate from the data of Ito and Kashiwagi [14]. We approximate using Equation (8.24): 4 ð70 kW=m2 Þ2 ð0:002 mÞ ¼ 12:5 kW2 =m3

compared to a measured value of 14 kW2 =m3 . This is probably the level of consistency to be expected for a ‘property’ such as . For unsteady lateral flame spread, Equation (8.25) can be applied in a quasi-steady fashion. We use the short-time approximation for the surface temperature (Equation (7.40)), T s T1

Table 8.1

00 q_ e FðtÞ ht

ð8:28Þ

Opposed flow properties for lateral flame spread on a vertical surface [16]

Material PMMA Fiberboard Plywood Wool carpet Polyurethane foam Paper on gypsum board

Tig ( C)

kc (kW/m2K)2s)

380 330 390 435 435 565

1.0 0.46 0.54 0.25 0.03 0.45

((kW)2/m3) 14.0 2.3 13.0 7.3 4.1 14.0

Ts;crit ( C) 20 210 120 335 215 425

205

EXPERIMENTAL CONSIDERATIONS FOR SOLID SURFACE SPREAD

where 2ht FðtÞ ¼ pﬃﬃﬃ

rﬃﬃﬃﬃﬃﬃﬃﬃﬃ kcp 1 t

Then Equation (8.25) can be expressed as v1=2 p

kcp h2t

1=2 h

00

q_ 00ig;crit q_ e FðtÞ

i

ð8:29Þ

Examine data taken under a natural convection lateral vertical surface spread for a particle board as shown in Figures 8.12(a) and (b) [17]. The correlation of these data based on Equation (8.29) suggests that the critical flux for ignition is 00

q_ ig;crit 16 kW=m2 and for the flame spread is 00

q_ s;crit 5 kW=m2 For an average ht ¼ 42 W=m2 K and T1 ¼ 25 C, we obtain Tig ¼ 25 þ

16 ¼ 406 C 0:042

and Ts;min ¼ 25 þ

5 ¼ 144 C 0:042

From the slope of the data in Figure 8.12(b),one can estimate kcp 1=2 ¼ 240 ðmm sÞ3=2 =J h2t Using ht ¼ 42 W=m2 K and kcp ¼ 0:93 (kW/m2 K)2 s (Table 7.4), can be estimated: ¼ ¼

kcp h2t ð240Þ2 0:93 ðkW=m2 KÞ2 s ð0:042 kW=m2 KÞ2 ½240 ðmm sÞ3=2 =J ð103 m=mmÞ3=2 ð103 J=kJÞ 2

¼ 9:15 kW2 =m3 This compares to a of 5 kW2 =m3 for another particle board in Table 7.4. The pseudo-property is a parameter that represents flow phenomena in addition to thermal and chemical properties of the material. Incidently, can be utilized for

206

FIRE SPREAD ON SURFACES AND THROUGH SOLID MEDIA

Figure 8.12

(a) Flame front movement for a wood particle board under opposed flow spread in still air under various external heating distributions. Symbols t, u, v, w are for downward spread; all the rest are for lateral spread on a vertical sample [17]. (b) Correlation based on Equation (8.29)[17]

207

EXPERIMENTAL CONSIDERATIONS FOR SOLID SURFACE SPREAD

thermally thin opposed flow spread as well as wind-aided. However, under wind-aided 00 conditions, it is not practical to link q_ f and f into a function. For these flow conditions 00 q_ f and f are distinctly independent variables.

8.5.2

Wind-aided

For turbulent wall fires, from the work of Ahmad and Faeth [18], it can be shown that 00

q_ f ¼ function ½xp ; YO2 ;1 ; Gr; Pr; fuel properties; flame radiation ðXr Þ; where xp ¼ length of the pyrolyzing region YO2 ;1 ¼ ambient oxygen mass fraction 2 Gr ¼ Grashof number ¼ g cos x3p = 1 ; Pr ¼ Prandtl number¼ 1 = 1 ¼ wall orientation angle from the vertical Xr ¼ flame of a radiation fraction The radiation dependence has been simplified using only Xr, but in general it is quite complex and significant for tall wall flames (> 1 m). The corresponding flame length can be shown to be [19] xf ¼ xp þ f ¼

1:02 ‘c Yo2;1 ð1 Xr Þ1=3

ð8:30aÞ

where a convective plume length scale is ‘c ¼

2=3 00 m_ F hc xp pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ 1 cp T1 g cos

ð8:30bÞ

The orientation effect included as cos is inferred from boundary layer theory and is not appropriate for an upward-facing surface at large where the boundary layer is likely to separate. It is more likely to apply to flame lengths on a bottom surface and not to lifted flames on a top surface. Similar results apply to laminar wall flames, but these correspond to Gr < 0:5 108 or about xp < 100 mm. However, irregular edges or rough spots can trip the flow to turbulent for xp as low as 50 mm. Hence, accidental fires of any 00 significance will quickly become turbulent. The burning rate per unit area, m_ F , depends 00 on the same variables as q_ f above. The Ahmad and Faeth [18] data encompass alcohols saturated into an inert wall of xp 00 up to 150 mm and xf up to 450 mm. Typically, q_ f is roughly constant over the visible flame extension ðf Þ with values of between 20 and 30 kW/m2. The same behavior is seen 00 for the radiatively enhanced burning of solid materials – again showing q_ f values of 2 20–30 kW/m over f for xf up to 1.5 m. These data are shown in Figure 8.13. Such empirical results for the flame heat flux are useful for obtaining practical estimates for upward flame spread on a wall.

208

FIRE SPREAD ON SURFACES AND THROUGH SOLID MEDIA

Figure 8.13

Example 8.1

Wall heat flux distribution [20]

It is found that flame length on a vertical wall can be approximated as 00 yf ¼ ð0:01 m2 =kWÞðQ_ yp Þ

00

where Q_ is the energy release flux in kW/m2 of the burning wall. For thick polyurethane foam listed in Table 8.1 compute (a) the upward spread velocity assuming that the region initially ignited is 0.1 by 0.1 m high and (b) the lateral spread rate on the wall occurs after 220 C. 00

Take Q_ ¼ 425 kW/m2 for the foam burning in air at 25 C. The adjacent surface immediately outside the spreading flame is heated to 225 C. Solution (a) We adopt y as the vertical coordinate, as shown in Figure 8.14. Therefore Equation (8.23) becomes 00

dyp ðq_ f Þ2 ðyf yp Þ yf yp ¼ ¼ 2 dt tf =4kcp ðTig Ts Þ

EXPERIMENTAL CONSIDERATIONS FOR SOLID SURFACE SPREAD

Figure 8.14

Example 8.1

We are given 00 yf ¼ 0:01Q_ yp

¼ ð0:01Þð425Þyp ¼ 4:25yp Based on Figure 8.13, we estimate 00

q_ f 25 kW=m2 Then we can compute

Tig Ts 2 tf ¼ kcp 00 4 q_ f ¼

ð0:03 ðkW=m2 KÞ2 sÞ 4

ð435 225ÞK 2 25 kW=m2

¼ 1:66s Substituting into the differential equation, we obtain

dyp 3:25 yp ¼ ¼ 1:95 yp dt 1:66 or solving with t ¼ 0, yp ¼ 0:1 m,

yp ðmÞ ¼ 0:1 e1:95 tðsÞ The speed is

vp ¼

dyp ¼ 0:195 e1:95 t dt

209

210

FIRE SPREAD ON SURFACES AND THROUGH SOLID MEDIA In 2 s, yp ¼ 4:9 m, and vp ¼ 9:67 m/s. U pward spread is very fast. However, for the empirical result, 00 yf 0:01 Q_ yp 00 only Q_ > 100 kW/m2 will allow spread to occur. The energy release flux of a material is a very critical variable for upward and wind-aided flame spread in general.

(b) The lateral spread is constant and given by up ¼ ¼

kcp ðTig Ts Þ2 4:1 ðkW=mÞ2 ð0:03 ðkW=m2 KÞ2 sÞð435 225Þ2

¼ 0:003 m=s ¼ 3 mm=s This is much, much slower than the upward flame speed and presents a significantly lower relative hazard.

8.6 Some Fundamental Results for Surface Spread deRis [4] was the first to obtain a complete theoretical solution for opposed flame spread on a surface. His solution is too detailed to present here, but we can heuristically derive the same exact results from our formulations for the thermally thin and thick cases. Consider Figure 8.15. We estimate the heat transfer lengths as g f BL

g u1

ð8:31Þ

where BL is the laminar boundary layer thickness. These are order of magnitude estimates only. The effects of chemical kinetics were ignored in deRis’ solution and only become a factor if the time for the chemical reaction is longer than the time for a gas particle to traverse g . By our order of magnitude estimates this flow time can be represented as BL =u1 . The ratio of the flow to chemical time is called the Damkohler number, Da, as defined in Equations (4.23) and (4.24). Using Equation (8.31), we can express this as Damkohler number; Da ¼

tFLOW BL =u1 ¼ tREACTION cp T1 =½E=ðRT1 Þ hc AeE=ðRT1 Þ

Figure 8.15

The deRis model

SOME FUNDAMENTAL RESULTS FOR SURFACE SPREAD

211

or Da ¼

g Ahc E eE=ðRT1 Þ u21 cp T1 RT1

ð8:32Þ

where g , , cp are gas phase properties. For small Da we expect gas phase kinetics to be important and control the spread process. For large Da we expect heat transfer to control the process. The latter corresponds to the deRis solution and our simple thermal model. We shall examine that aspect. From Equation (8.31), the flame heat flux is estimated by 00

q_ f ¼ kg

Tf Tig g

ð8:33Þ

with Tf , the flame temperature. Combining and substituting into Equation (8.6) for the thermally thin case gives vp ¼ If f ¼

½kg ðTf Tig Þ=g f cp dðTig Ts Þ

pﬃﬃﬃ 2g , we obtain exactly the deRis solution for the thermally thin case, pﬃﬃﬃ 2kg ðTf Tig Þ vp ¼ cp dðTig Ts Þ

ð8:34Þ

In the thermally thick case (Equation (8.21)), vp ¼

½kg ðTf Tig Þ=g 2 f ðkcp ÞðTig Ts Þ2

¼

ðkg2 u1 = g ÞðTf Tig Þ2 ðkcp ÞðTig Ts Þ2

or vp ¼

ðkcp Þg u1 ðTf Tig Þ2 ðkcp ÞðTig Ts Þ2

ð8:35Þ

which is exactly the theoretical result of deRis [4]. The deRis solutions were examined experimentally by Fernandez-Pello, Ray and Glassman [5,6]. Results are shown for thick PMMA in Figure 8.16 [6] as a function of u1 and YO2 ;1 . It should be recognized that Tf depends on YO2 ;1 , the free stream oxygen mass fraction, and A, the pre-exponential factor, depends on YF;s and YO2 ;1 . These experimental results show that for the rise and fall of the data for large u1 , the deRis model is not followed. This is because the chemical reaction times are becoming relatively large, Da ! small. Indeed, the results can be correlated by vp ðkcp ÞðTig Ts Þ2 u1 ðkcp Þg ðTf Tig Þ2

¼ f ðDaÞ ! 1 as Da ! large

ð8:36Þ

212

FIRE SPREAD ON SURFACES AND THROUGH SOLID MEDIA

Figure 8.16

Flame spread rate over thick PMMA sheets as a function of the opposed forced flow velocity for several flow oxygen mass fractions (Fernandez-Pello, Ray and Glassman [6])

where f ðDaÞ is monotonic and asymptotic to 1 as Da becomes large [6]. The flame temperature can be approximated by Tf Tig ¼

YO2 ;1 hc cp;g r

ð8:37Þ

Loh and Fernandez-Pello [8] have shown that the form of Equation (8.30) appears to hold for laminar wind-aided spread over a flat plate. Data are shown plotted in Figure 8.17 and are consistent with vp u1 YO2 2 ;1 following Equations (8.36) and (8.37). Note, from Figures 8.16 and 8.17 that at u1 ¼ 1 m=s and YO2 ;1 ¼ 1; vp 0:6 cm=s for opposed flow and vp 0:6 m=s for wind-aided spread

213

EXAMPLES OF OTHER FLAME SPREAD CONDITIONS

Figure 8.17

Flame spread rate over thick PMMA sheets as a function of the concurrent forced flow velocity for several flow oxygen mass fractions (Loh and Fernandez-Pello [8])

8.7 Examples of Other Flame Spread Conditions Many fire spread problems can be formulated by the deceptively simple expression Flame speedðvp Þ ¼

flame heated length ðf Þ time to ignite ðtf Þ

The key to a successful derivation of a particular flame spread phenomena is to establish the mechanism controlling the spread in terms of expressions for f and tf . We will not continue to develop such approximate expressions, but will show some results that are grounded in data and empirical analyses.

8.7.1

Orientation effects

Figure 8.18 shows spread rates measured for thin films ( 0.2 mm) of polyethylene teraphthalate (PET) and paper laid on fiberglass at various orientations spreading up and down [19]. The results can partly be explained by relating the momentum to buoyancy as u21 ðg cos Þ BL

ð8:38Þ

214

FIRE SPREAD ON SURFACES AND THROUGH SOLID MEDIA

Figure 8.18

Spread as a function of orientation and angle [19]

and from Equation (8.31) u1 ðg g cos Þ1=3

ð8:39Þ

Thus, we can replace u1 in Equation (8.36) and apply it to both opposed and wind-aided cases. For upward or wind-aided spread the speed increases as cos increases to the vertical orientation. For downward or opposed flow spread, the speed is not significantly affected by changes in until the horizontal inclination is approached for the bottom orientation ð90 60Þ. In that region, the downward spread becomes wind-aided as a stagnation plane flow results from the bottom. Figure 8.19 gives sketches of the

Figure 8.19

Flame shapes during spread for Figure 8.18 [19]

EXAMPLES OF OTHER FLAME SPREAD CONDITIONS

215

flame shapes for spread under some of these orientations. For spread on the bottom surface we always have a relatively large f . For spread on the top, we get ‘lift-off’ of the flame between ¼ 30 to 60 .

8.7.2

Porous media

Thomas [21,22] has analyzed and correlated data for porous, woody material representative of forest debris, grass fields, cribs and even blocks of wooden houses. He finds that vp u1 =b for the wind-aided case, and vp 1 b for natural convection, where b is the bulk density of the fuel burned in the array. Figure 8.20 shows the wind-aided data and Figure 8.21 is the no-wind case. From Thomas [21,22], an empirical formula can be shown to approximate much of the data:

b vp ¼ cð1 þ u1 Þ in m; kg and s units

ð8:40Þ

where c can range from about 0.05 for wood cribs ðb 10--100 kg=m3 Þ up to 0.10 for forest fuels ðb 1--5 kg=m3 Þ. The urban house array data were represented as 10 kg=m3 in Figure 8.20.

Figure 8.20

Wind-aided fire spread through porous arrays as a function of wind speed [21]

216

FIRE SPREAD ON SURFACES AND THROUGH SOLID MEDIA

Figure 8.21

8.7.3

Natural convection fire spread through a porous fuel bed from Thomas [22]

Liquid flame spread

Flame spread over a deep pool of liquid fuel depends on flow effects in the liquid. These flow effects constitute: (a) a surface tension induced flow since surface tension ðÞ is decreasing with temperature (T) and (b) buoyancy induced flow below the surface. For shallow pools, viscous and heat transfer effects with the floor will be factors. Figure 8.22 shows these mechanisms. To ignite the liquid, we must raise it to its flashpoint. Then the spread flame with its energy transport mechanisms must heat the liquid at temperature T to its flashpoint, TL . Data from Akita [23] in Figure 8.23 show the effect of increasing bulk liquid temperature (T) on the spread rate. At T ¼ 10 C for the methanol liquid, surface tension and viscous effects begin to cause a combustible mixture to form ahead of the advancing flame and its ignition causes a sudden increase in the flame speed. This process is repeated as a pulsating flame until T ¼ 1 C, where the speed again becomes steady. After T ¼ 11 C, the flashpoint, the flame spread is no longer a surface

Figure 8.22

Mechanism for flame spread on a liquid

EXAMPLES OF OTHER FLAME SPREAD CONDITIONS

Figure 8.23

217

Relationship between the liquid temperature and the rate of plane flame spread of methanol in a vessel 2.6 cm wide and 1.0 cm deep [23]

phenomenon. Now it is flame propagation through the premixed fuel and air in the gas phase. Increases in speed continue with T up to the liquid temperature corresponding to stoichiometric conditions (20.5 C), after which the mixture becomes fuel rich at a constant speed.

8.7.4

Fire spread through a dwelling

An interesting correlation of the rate of fire growth measured by flame volume was given by Labes [24] and Waterman [25]. They show that after room flashover in a room of volume V0 , the fire volume ðVf Þ grows as dVf ¼ k f Vf dt

ð8:41Þ

where kf depends on the structure, its contents and ventilation. Two figures (Figures 8.24 and 8.25) display data taken from two dwellings. A simple explanation of Equation (8.41) can be formulated by considering fire spread illustrated for the control volume in Figure 8.26. We write a conservation of energy for the CV as 00

cp vp AðTig T1 Þ ¼ q_ f A:

ð8:42Þ

The flame volumetric rate of change is related to flame spread as dVf ¼ vp A dt

ð8:43Þ

218

FIRE SPREAD ON SURFACES AND THROUGH SOLID MEDIA

Figure 8.24

Correlation of volumetric fire growth for a dwelling with similar compartment characteristics [24]

for constant A. The heat flux for a ‘thin’ flame model can be represented as 00

q_ f ¼ f Tf4

ð8:44Þ

with f ¼ 1 ef ðVf =AÞ f

Figure 8.25

Vf A

ð8:45Þ

Correlation of volumetric fire growth for a dwelling under changing conditions [25]

REFERENCES

Figure 8.26

219

Fire spread in a dwelling after room flashover

where f is a radiative absorption coefficient for the flame. Combining yields dVf f Tf4 ¼ Vf dt cp ðTig T1 Þ

ð8:46Þ

or the form of Equation (8.41), the experimental correlation. It is interesting to note that the fire doubling time is constant at about 7.5 minutes for dwelling, having similar compartments in Figure 8.24. Doubling times could range from roughly 1 to 10 minutes over a typical spectrum of dwellings and wind conditions.

References 1. Williams, F.A., Mechanisms of fire spread, Proc. Comb. Inst., 1976, 16, p. 1281. 2. Fernandez-Pello, A.C., The solid phase, in Combustion Fundamentals of Fire (ed. G. Cox), Academic Press, London, 1995. 3. Thomas, P.H., The growth of fire – ignition to full involvement, in Combustion Fundamentals of Fire (ed. G. Cox), Academic Press, London, 1995. 4. deRis, J.N., Spread of a laminar diffusion flame, Proc. Comb. Inst., 1969, pp. 241–52. 5. Fernandez-Pello, A.C., Ray, S.R. and Glassman, I., A study of heat transfer mechanisms in horizontal flame propagation, J. Heat Transfer, 1980, 102(2), 357–63. 6. Fernandez-Pello, A.C., Ray, S.R. and Glassman, I., Flame spread in an opposed forced flow: the effect of ambient oxygen concentration, Proc. Comb. Inst., 1980, 18, pp. 579–89. 7. Altenkirch, R.A., Eichhorn, R. and Shang, P.C., Buoyancy effects on flames spreading down thermally thin fuels, Combustion and Flame, 1980, 37, 71–83. 8. Loh, H.T. and Fernandez-Pello, A.C., A study of the controlling mechanisms of flow assisted flame spread, Proc. Comb. Inst., 1984, 20, pp. 1575–82. 9. Loh, H.T. and Fernandez-Pello, A.C., Flow assisted flame spread over thermally thin fuels, in Proceedings of 1st Symposium on Fire Safety Science, Hemisphere, New York, 1986, p. 65. 10. Sibulkin, M. and Kim, J., The dependence of flame propagation on surface heat transfer. II. Upward burning, J. Combust. Sci. Technol., 1977, 17, 39. 11. Rothermel, R.C., Modeling fire behavior, in 1st International Conference on Forest Fire Research, Coimbra, Portugal, November 1990.

220

FIRE SPREAD ON SURFACES AND THROUGH SOLID MEDIA

12. Long Jr, R.T., An evaluation of the lateral ignition and flame spread test for material flammability assessment for micro-gravity environments, MS Thesis, Department of Fire Protection Engineering, University of Maryland, College Park, Maryland, 1998. 13. Saito, K., Williams, F.A., Wichman, I.S. and Quintiere, J.G., Upward turbulent flame spread on wood under external radiation, Trans. ASME, J. Heat Transfer, May 1989, 111, 438–45. 14. Ito, A. and Kashiwagi, T., Characterization of flame spread over PMMA using holographic interferometry sampling orientation effects, Combustion and Flame, 1988, 71, 189. 15. ASTM E-1321-90, Standard Method for Determining Material Ignition and Flame Spread Properties, American Society for Testing and Materials, Philadelphia, Pennsylvania, 1990. 16. Quintiere, J.G. and Harkleroad, M., New Concepts for Measuring Flame Spread Properties, NBSIR 84-2943, National Bureau of Standards, Gaithersburg, Maryland, 1984. 17. Quintiere, J.G., Harkleroad, M. and Walton, D., Measurement of material flame spread properties, J. Combust. Sci. Technol., 1983, 32, 67–89. 18. Ahmad, T. and Faeth, G.M., Turbulent wall fires, in 17th Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, Pennsylvania, Proc. Comb. Inst., 1979, 17, p. 1149–1162. 19. Quintiere, J.G., The effects of angular orientation on flame spread over thin materials, Fire Safety J., 2001, 36(3), 291–312. 20. Quintiere, J.G., Harkleroad, M. and Hasemi, Y., Wall flames and implications for upward flame spread, Combust. Sci. Technol., 1986, 48, 191–222. 21. Thomas, P.H., Rates of spread of some wind driven fires, Forestry, 1971, 44(2), 155–7. 22. Thomas, P.H., Some aspects of growth and spread of fires in the open, Forestry, 1967, 40(2), 139–64. 23. Akita, K., Some problems of flame spread along a liquid surface, in 14th Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, Pennsylvania, Proc. Comb. Inst., 1975, 14, p. 1075. 24. Labes, W.G., The Ellis Parkway and Gray Dwelling burns, Fire Technol., May 1966, 2(4), 287. 25. Waterman, T.E., Experimental Structure Fires, IITRI Final Technical Report J6269, Chicago, Ilinois, July 1974.

Problems 8.1

A thin vertical drapery is ignited. The flame spreads up and down both sides. Note that the axis of symmetry is equivalent to an insulated boundary. The properties of the drape are given below:

T1 ¼ 25 C Yox;1 ¼ 0:233 k ¼ 0:08 W=m K c ¼ 1:5 J=g K ¼ 100 kg=m3 Tig ¼ 400 C To ¼ 25 C, initial temperature Thickness ¼ 2 mm

Consider the analysis in two dimensions only, i.e. per unit depth into the page.

PROBLEMS

221

(a) Calculate the flame temperature from thin diffusion flame theory, using cp;g ¼ 1 J=g K and kg ¼ 0:03 W/m K (b) Calculate the downward spread flame speed. (c) Calculate the initial upward spread flame speed if the flame extension ¼ axp , where xp is the pyrolysis length and a ¼ 1. The heat flux of the flame is 4 W/cm2 and the initial pyrolysis zone is 2 cm in length. Note that diffusion flame theory will show the flame temperature is approximately given by Tf T1 þ Yox;1 ðhox =cp Þ, where hox is the heat of combustion per unit mass of oxygen consumed. 8.2

A 1 m high slab of PMMA is ignited at the bottom and spreads upward. Ignition is applied over a length of 5 cm. If the slab is 1 mm thick and ignited on both sides, then how long will it 0 take to be fully involved in pyrolysis? Assume the flame height is governed by xf ﬃ 0:01Q_ , 00 00 where xf is in meters and Q_ 0 is in kW/m. The flame heat flux is 25 kW/m2; assume m_ ¼ q_ =L, where L ¼ 1:6 kJ/g. Let T1 ¼ 20 C and obtain data for PMMA from Tables 2.3, 7.5, 7.6 and 8.1.

8.3

Re-work Problem 8.2 using a PMMA thickness of 50 mm.

8.4

If the PMMA in Problem 8.2 is ignited at the top for the thick case, compute the time for full involvement. Let ¼ 14 (kW)2/m3.

8.5

A thick wood slab spread a flame downward at 0.1 mm/s; the slab was initially at 20 C. If the slab were preheated to 100 C, what is the spread rate? Assume Tig ¼450 C.

8.6

A paper match once ignited releases 250 kW/m2 of energy and burns for 60 seconds. The match is a thin flat configuration (0.5 mm thick), 0.5 cm wide and 4 cm long (see the figure below). The person striking the match grips it 1 cm from the top and inadvertently turns it upside down. Ignore edge effects.

1 cm 4 cm

0.5 cm x 0.5cm

The initial region ignited is 0.5 cm and the ambient temperature is 20 C. The flame height 0 0 during flame spread is given by xf ¼ ð0:01m2 =kWÞQ_ , where Q_ is the energy release rate per unit width in kW/m.

222

FIRE SPREAD ON SURFACES AND THROUGH SOLID MEDIA Match properties: Density ¼ 100 kg/m3 Specific heat ¼ 1:8 J/g K Ignition temperature ¼ 400 C Flame heat flux ¼ 25 kW/m2 (a) If this match material is exposed to a heat flux of 25 kW/m2 and insulated on the opposite face, how long will it take to ignite? (b) For the finger scenario described, determine the pyrolysis position just when the flame tip reaches the finger tip. (c) How long will it take the flame to reach the fingertip?

8.7

A thick material is tested for ignition and flame spread.

4W/cm2

v

(a) An ignition test is conducted under a radiative heat flux of 4 W/cm2. The initial temperature is 20 C. It is found that the surface temperature rises to 100 C in 3 seconds. The ignition (temperature?) is 420 C. (i) Estimate the property kc of the material. (ii) When will it ignite? (b) Two flame spread experiments are conducted to examine downward spread velocity. The material is preheated in an oven to 100 C and to 200 C respectively for the two experiments. Calculate the ratio of spread rates for the two experiments ðv100 C =v200 C Þ. 8.8

A thick polystyrene wainscotting covers the wall of a room up to 1 m from the floor. It is ignited over a 0.2 m region and begins to spread. Assume that the resulting smoke layer in the room does not descend below 1 m and no mixing occurs between the smoke layer and the lower limit. The initial temperature is 20 C, the ambient oxygen mass fraction is 0.233 and the specific heat of air is 1 J/g K. Polystyrene properties: Density ðÞ ¼ 40 kg/m3 Specific heat ðcp Þ ¼ 1.5 J/g K

PROBLEMS

223

Conductivity ðkÞ ¼ 0:4 W/m K Vaporization ðTv Þ and ignition temperature ðTig Þ are both equal to 400 C Heat of combustion ðhc Þ ¼ 39 kJ/g Stoichiometric O2 to fuel ratio (r) ¼ 3 g/g Heat of gasification ðLÞ ¼ 1.8 kJ/g

Smoke

1m wainscotting 0.2 m ignition

Calculate the upward spread velocity at 0.5 m from the floor. The flame height from the floor is 1.8 m and the heat flux from the flame is estimated at 3 W/cm2. 8.9

A wall is ignited by a spill of gasoline along the wall–floor interface. The width of the spill is 0.5 m. The energy release rate of the gasoline is 300 kW/m (width). In all calculations, the dominant heat flux is from the flame to the wall. Only consider this heat flux; it is 30 kW/m2 for all flames. The initial temperature is 20 C. The relationship between the wall flame height 0 and the energy release rate per unit width ðQ_ Þ is 0 yf ðmÞ ¼ 0:01Q_ ðkW=mÞ

The properties of the wall material are: Tig ¼ 350 C Ts;min ¼20 C hc ¼25 kJ/g ¼ 15 kW2/m3 L ¼1.8 kJ/g kc ¼ 0:45 (kW/m2 K)2 s (a) When does the wall ignite after the gasoline fire starts, and what is the region ignited? (b) What is the maximum width of the wall fire 60 s after ignition of the wall? (c) What is the flame height 60 s after the wall ignition? (d) What is the maximum height of the pyrolysis front after 60 s from wall ignition? (e) What is the energy release rate of the wall 60 s after its ignition? (f) Sixty seconds after ignition of the wall, the gasoline flame ceases. The wall continues to burn and spread. What is the maximum pyrolysis height 30 s from the time the gasoline flame ceases?

224

FIRE SPREAD ON SURFACES AND THROUGH SOLID MEDIA

8.10

Flame spread is initiated on a paneled wall in the corner of a family room. Later, a large picture window, opposite the corner, breaks. A plush thick carpet covers the floor and a large sofa is in front of the picture window. The ceiling is noncombustible plaster. Explain the most probable scenario for causing the window to break and shatter.

8.11

A vat of flammable liquid ignites and produces a flame that impinges on a 1 m radius of the paper-facing insulation on the ceiling. The flame temperature is 500 oC and convectively heats the ceiling with a heat transfer coefficient of 50 W/m2 K. The initial temperature is 35 oC. The paper-face on the ceiling is 1 mm thick with properties as shown:

ro = 1 m

Ignition temperature, Tig ¼ 350 C Density, ¼ 250 kg/m3 Specific heat, c ¼ 2100 J/kg K (a) Compute the time to ignite the paper.

(b) Compute the initial flame spread rate under the ceiling. The flame stretches out from the ignited region by 0.5 m and applies a heat flux of 30 kW/m2 to the surface.

ro = 0.5 m

PROBLEMS 8.12

8.13

225

Flame spread into the wind can depend on (check the best answer): (a) Wind speed

Yes —

Not necessarily —

No —

(b) Flame temperature

—

—

—

(c) Flame length in the wind direction

—

—

—

(d) Pyrolysis region heat flux

—

—

—

The temperature at the edge of a thick wood element exposed to a uniform heat flux can be expressed, in terms of time, by the approximate equation given below for small time: 4:2 00 pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ Ts T1 pﬃﬃﬃ q_ e t=kc The notation is consistent with that of the text. The properties of the wood are: Thermal conductivity Density Specific heat Piloted ignition temperature Ambient temperature

0.12 W/m K 510 kg/m3 1380 J/kg K 385 C 25 C

After ignition of the edge, the external heat flux is removed and it is no longer felt by the wood, in any way. A researcher measures the heat flux of the flame, ahead of the burning wood, as it spreads steadily and horizontally along the edge. That measurement is depicted in the plot below. Compute the flame speed on the wood, based on this measurement.

x

60 Heat flux kW/m2

0

8.14

1

2

3

x (mm)

The ASTM Steiner Tunnel Test ignites a material over 1.5 m at its burner with a combined flame length of 3.2 m from the start of the ignition region. As the flame moves away from the burner influence, its burning rate becomes steady over a 0.3 m length and flame extends only 0.2 m beyond this burning length. The flame heat flux remains constant at 25 kW/m2. Ignore any radiant feedback and increases in the tunnel air temperature. What is the ratio of steady speed to its initial speed induced by the burner? The material is considered to be thick.

9 Burning Rate

9.1 Introduction Burning rate is strictly defined as the mass rate of fuel consumed by the chemical reaction, usually but not exclusively in the gas-phase. For the flaming combustion of solids and liquids, the burning rate is loosely used to mean the mass loss rate of the condensed phase fuel. However, these two quantities – mass loss rate and burning rate – are not necessarily equal. In general, ½Fuel mass loss rate ¼ ½fuel burning rate þ ½rate of inert gases released with the fuel þ ½rate of fuel gases ðand sootÞ not burned in the flame or m_ F ¼ m_ F;R þ m_ F;I þ m_ F;U

ð9:1Þ

In this chapter we shall consider only m_ F ¼ m_ F;R

ð9:2Þ

assuming all of the fuel is burned and there are no inerts in the evolved fuel. The latter assumption could be relaxed since YF;o (the mass concentration of fuel in the condensed phase) will be a parameter, and in general YF;o 1. In reality, all fuels will have unburned products of combustion to some degree. This will implicitly be taken into account since chemical properties for fuels under natural fire conditions are based on measurements of the fuel mass loss rate. Hence, in this chapter, and hereafter for fire

Fundamentals of Fire Phenomena James G. Quintiere # 2006 John Wiley & Sons, Ltd ISBN: 0-470-09113-4

228

BURNING RATE

conditions, we regard the heat of combustion as the chemical rate of energy released (firepower) per unit fuel mass loss, hc

Q_ chem m_ F

ð9:3Þ

Similarly, yields of species, yi , are given as yi

m_ i;chem m_ F

ð9:4Þ

where m_ i;chem is the mass rate of production of species i from the fire. Factors that influence the burning rate of ‘fuel packages’ include (1) the net heat flux received at the surface, q_ 00 , and (2) the evaporative or decomposition relationships for the fuel. We can represent each as (1) Net heat flux: q_ 00 ¼ flame convective heat flux;

q_ 00f;c

þ flame radiative heat flux;

q_ 00f;r

þ external environmental radiative heat flux;

surface radiative heat loss;

q_ 00e

Ts4

ð9:5Þ

. (2) Fuel volatilization mass flux: m_ 00F ¼ function ðchemical decomposition kinetics; heat conduction; charring and possibly other physical; chemical and thermodynamic propertiesÞ Some of the thermal and fluid factors affecting the burning rate are illustrated in Figure 9.1. Convective heat flux depends on the flow conditions and we see that both

Figure 9.1

Factors influencing the burning rate

INTRODUCTION

Figure 9.2

229

Idealized burning rate behavior of thick charring and noncharring materials: (a) mass loss behaviour and (b) temperature behavior

boundary layer and plume flows are possible for a burning fuel package. For flames of significance thickness (e.g. 5 cm, then f 0:05), flame radiation will be important and can dominate the flame heating. Since the particular flow field controls the flame shape, flame radiation will depend on flow as well as fire size (m_ F ). The external radiative heating could come from a heated room subject to a fire; then it will vary with time, or it could come from a fixed heat source of a test, where it might vary with distance. The radiative loss term depends on surface temperature, and this may be approximated by a constant for liquids or noncharring solids. For liquids, this surface temperature is nearly the boiling point, and will be taken as such. For charring materials the surface temperature will increase as the char layer grows and insulates the material. It can range from 500 to 800 C depending on heat flux ð q_ 00r =Þ1=4 or char oxidation. The condensed phase properties that affect fuel volatilization are complex and the most difficult to characterize for materials and products, in general. The mass loss rate depends on time as well as the imposed surface heat flux. Figure 9.2 illustrates idealized experimental data for the mass loss rate of charring and noncharring materials. Figures 9.3 and 9.4 give actual data for these material classes along with surface temperature information. Initially, both types of materials act in the same way since a thin char layer has little effect. However, at some char thickness, the mass loss rate abruptly begins to decrease for the charring material as the effect of char layer thickness becomes important. The decreasing behavior in burning rate roughly follows t 1=2 . This behavior

230

BURNING RATE

Figure 9.3 (a) Transient mass loss rate for polyethylene exposed to 36 kW/m2. (b) surface temperature results for polyethylene with 36 kw/m2 irradiance. (c) Transient mass loss rate exposed to 70 kw/m2 irradiance [1]

can be derived from considering a thermal penetration depth that is defined by a fixed temperature,pindicative of the onset of charring, Tchar . For a pure conduction model, we ﬃﬃﬃﬃﬃ pﬃﬃﬃﬃﬃﬃﬃ know T t so we would expect c c t where c is the char diffusivity. By a mass balance for the control volume (Figure 9.2b), we approximate d ðc c Þ þ m_ 00F o w ¼ 0 dt 00 where w ¼ dc =dtpand ﬃﬃﬃﬃﬃﬃﬃﬃﬃ o ¼ c þ volatiles . Therefore, m_ F ¼ ðo c Þðdc =dtÞ or 00 m_ F ½ðo c Þ=2 c =t, where t is the time after reaching the peak. In contrast, the noncharring ‘liquid evaporative-type’ material reaches a steady peak burning rate. For both types of materials, finite thickness effects and their backface boundary condition will affect the shape of these curves late in time. This is seen by the increased burning rate of the 2.5 cm thick polypropylene at 70 W/m2 in Figure 9.3(c), but not for the 5 cm thick red oak material at 75 kW/m2 in Figure 9.4(a). A crude, but effective way to deal with the complexities of transient burning and fuel package properties is to consider liquid-like, quasi-steady burning. By Equation (6.34),

INTRODUCTION

Figure 9.4

231

(a) Burning rate of red oak at 75 kW/m2 (b) Temperatures measured for red oak at 75 kW/m2 [2]

we derived for the steady burning of a liquid that m_ 00F ¼

q_ 00 L

ð9:6Þ

where here we call L an effective heat of gasification. This overall property can be determined from the slope ð1=LÞ of peak m_ 00F versus q_ 00 data. Such data are shown in Figure 9.5 for polypropylene burning a 10 cm square horizontal sample under increasing radiant heat flux [1]. Thus, L ¼ 3:07 kJ=g for this polypropylene specimen. For peak data on Douglas fir in Figure 9.6, the best fit lines give L of 6.8 kJ/g for heating along the grain and 12.5 kJ/g across the grain. It is easier to volatilize in the heated along-the-grain orientation, since the wood fibers and nutrient transport paths are in this heated direction (see Figure 7.10). While the L representation for noncharring materials may be very valid under steady burning, the L for a charring material disguises a lot of phenomena. It is interesting to point out that a noncharring material might be regarded as having all fuel converted to vapor, whereas for a charring material only a fraction is

232

BURNING RATE

Figure 9.5

Peak mass loss rate of polypropylene [1]

converted. Hence an L-value, based on the original fuel mass. ðLo Þ, is related to the common value based on mass loss as Lo ðkJ=g originalÞ ¼ LðkJ=g volatilesÞð1 Xc Þ

ð9:7Þ

where Xc is the char fraction, the mass of char per unit mass of original material. The char fraction was measured for the Douglas fir specimen as 0:25 0:07 along and 0:45 0:15 across the grain orientations. Thus, Lo ¼

ð6:8Þð1:0 0:25Þ ¼ 5:1 kJ=g or ð12:5Þð1:0 0:45Þ ¼ 6:9 kJ=g

In this form, the energy needed to break the original polymer bonds to cause ‘unzipping’ or volatilization with char is closer to values representative of noncharring solid polymers. Table 9.1 gives some representative values found for the heats of gasification.

Figure 9.6

Peak burning rate as a function of the incident heat flux for Douglas fir [2]

DIFFUSIVE BURNING OF LIQUID FUELS

233

Table 9.1 Effective heats of gasification Material

Source

kJ/g mass lost

[1] [3] [1] [3] [1] [3] [1] [3] [3] [4] [3] [3] [4] [2] [2] [2] [2]

3.6 1.8, 2.3 3.1 2.0 3.8 2.4 2.8 1.6 1.3–1.9 4.0, 7.3 1.2–2.7 1.2 5.6 12.5,(6.8)a 9.4,(4.6) 9.4,(7.9) 4.7,(6.3)

Polyethylene Polyethylene Polypropylene Polypropylene Nylon Nylon Polymethylmethacrylate Polymethylmethacrylate Polystyrene Polystyrene Polyurethane foam, flexible Polyurethane foam, rigid Polyurethane foam, rigid Douglas fir Redwood Red oak Maple a

Heat transfer parallel to the wood grain.

It should be realized that the use of L values from Table 9.1 must be taken as representative for a given material. The values can vary, not just due to the issues associated with the interpretation of experimental data, but also because the materials listed are commercial products and are subject to manufacturing and environmental factors such as purety, moisture, grain orientation, aging, etc. We shall formalize our use of L by considering analyses for the steady burning of liquid fuels where the heat of gasification is a true fuel property.

9.2 Diffusive Burning of Liquid Fuels 9.2.1

Stagnant layer

In the study of steady evaporation of liquid, we derived in Equation (6.34), and given as Equation (9.6), m_ 00F ¼ where L ¼ hfg þ

R Ts Ti

q_ 00 L

cp;l dT with

hfg ¼ energy required to vaporize a unit mass of liquid at temperature Ts R Ts Ti cp;l dT ¼ energy needed to raise a unit mass of liquid from its original temperature Ti to the vaporization temperature on the surface, Ts . This latter energy is sometimes referred to as the ‘sensible energy’, since it is realized by a temperature change

234

BURNING RATE

Figure 9.7

Stagnant layer model – pure convective burning

We shall consider at first only steady burning. Moreover, we shall restrict our heating to only that of convection received from the gas phase. This is flame heating. Therefore, purely convective heating is given as dT 00 q_ ¼ k ¼ m_ 00F L ð9:8Þ dy y¼0 where k is the gas phase thermal conductivity; the coordinate system is shown in Figure 9.7. A special approximation is made for the gas- phase equations which is known as the stagnant layer model. The region of interest is the control volume depicted. It is a planar region of width x and thickness . Here can be imagined to represent the fluiddynamical boundary layer where all of the diffusional changes of heat, mass and momentum are occurring. In Figure 9.7, we illustrate such changes by the portrayal of the temperature, fuel and oxygen concentration profiles. Combustion is occurring within this boundary layer. The rate of combustion can be governed by an Arrhenius law m_ F 000 ¼ AðYF ; YO2 Þe E=ðRTÞ

ð9:9Þ

for each point in the boundary layer with the pre-exponential parameter, A, a function of reactant concentrations and temperature. However, as depicted for this diffusion flame, the fuel and oxygen concentration may not penetrate much past each other. Indeed, if the chemical reaction time is very fast relative to the diffusion time (Equation (8.32)) tchem 1 u21 cp T ¼ ¼ Da Ahc ½E=ðRTÞe E=ðRTÞ tDiff or tchem u1 x2 2 cp T ¼ 2 tDiff x Ahc ½E=ðRTÞe E=ðRTÞ kT ¼ Re2x Pr 2 2 Ahc x ½E=ðRTÞe E=ðRTÞ

ð9:10Þ

DIFFUSIVE BURNING OF LIQUID FUELS

235

where Rex is the Reynolds number, u1 x= , and Pr is the Prandtl number, =. Alternatively, a chemical length can be deduced by inspection as chem

rﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ kT ¼ Ahc

ð9:11Þ

Accordingly, tchem Re2chem Pr 2 ¼ tDiff ½E=ðRTÞe E=ðRTÞ

ð9:12Þ

where Rechem u1 chem = , a chemical Reynolds number. The physical interpretation of chem is a region much thinner than , the boundary layer thickness, in which most of the chemical reaction is experienced. Outside of this region, A is nearly zero, while within the region A is relatively large. From our experience with the functions A and E=ðRTÞ, we expect and will require that tchem tDiff . In the flame region ðchem Þ, Rechem is small and we might expect that diffusion and viscous effects dominate the characteristics of this reaction region. Indeed, if the chemical time is approximated as zero, then chem would be an infinitesimal sheet – a flame sheet approximation. The concentration profiles of YO2 and YF would no longer overlap, but would be zero at the sheet. We will use this result. Since diffusional effects are most important, we wish to emphasize these processes in the gas phase. For the control volume selected in Figure 9.7, the bold assumption is made that transport processes across the lateral faces in the x direction do not change – or change very slowly. Thus we only consider changes in the y direction. This approximation is known as the stagnant layer model since the direct effect of the main flow velocity ðuÞ is not expressed. A differential control volume y x unity is selected. By a process of applying the conservation of mass, species and energy to the control volume ðy x unit lengthÞ, and expressing all variables at y þ y as f ðy þ yÞ f ðyÞ þ

df y þ OðyÞ2 dy

using Taylor’s expansion theorem, we arrive at the following: Conservation of mass d ðvÞ ¼ 0 dy

ð9:13Þ

d d dYi ðvYi Þ ¼ D þ m_ i 000 dy dy dy

ð9:14Þ

Conservation of species

where Fick’s law of diffusion has been used; m_ i 000 is the production rate per unit volume of species i.

236

BURNING RATE

Diffusion mass flux m_ 00i; Diff ¼ D

dYi dy

ð9:15Þ

Conservation of energy d d dT cp ðvTÞ ¼ k þ m_ F 000 hc dy dy dy

ð9:16Þ

with only heat transfer by conduction considered from Fourier’s law: q_ 00 ¼ k

dT dy

ð9:17Þ

where m_ F 000 is the gas phase fuel consumption rate per unit volume and hc is the heat of combustion. We have made the further assumptions: 1. Radiant heat transfer is ignored. 2. Specific heat is constant and equal for all species. 3. Only laminar transport processes are considered in the transverse y direction. Furthermore, we will take all other properties as constant and independent of temperature. Due to the high temperatures expected, these assumptions will not lead to accurate quantitative results unless we ultimately make some adjustments later. However, the solution to this stagnant layer with only pure conduction diffusion will display the correct features of a diffusion flame. Aspects of the solution can be taken as a guide and to give insight into the dynamics and interaction of fluid transport and combustion, even in complex turbulent unsteady flows. Incidentally, the conservation of momentum is implicitly used in the stagnant layer model since: Conservation of momentum dp ¼0 dy

ð9:18Þ

is an expression for the conservation of momentum provided is small. Recall for boundary layer flows that 1 n x Rex

ð9:19Þ

where n ¼ 12 for laminar flows and about 45 for turbulent flow. Under natural convection we can replace Rex by Grx ½ with Grx gx3 = 2 . Thus, we see for boundary layer flows that is small if Rex is large enough, i.e. Rex > 103 .

DIFFUSIVE BURNING OF LIQUID FUELS

9.2.2

237

Stagnant layer solution

The simple but representative chemical reaction is considered with one product, P, as 1 g FðfuelÞ þ r g O2 ðoxygenÞ ! ðr þ 1Þg PðproductÞ

ð9:20Þ

The inclusion of more products only makes more details for us with no change in the basic content of the results. Let us examine the integration of the governing differential equations. From Equation (9.13), Z y d ðvÞ ¼ 0 0 dy gives m_ 00F ¼ v ¼ constant

ð9:21Þ

Since at y ¼ 0, the mass flux v is the mass loss rate of burning rate evolved from the condensed phase. From Equation (9.18) we also realize that we have a constant pressure process pðyÞ ¼ p1

ð9:22Þ

The remaining energy and species equations can be summarized as dT d2 T

k 2 ¼ m_ F 000 hc dy dy 8 > m_ F 000 ; i ¼ F < 2 dY d Y i i

D 2 ¼

r m_ F 000 ; i ¼ O2 m_ 00F > dy dy : ðr þ 1Þm_ F 000 ; i ¼ P

cp m_ 00F

ð9:23Þ ð9:24aÞ ð9:24bÞ ð9:24cÞ

We have four equations but five unknowns. Although a constant, in this steady state case, m_ 00F is not known. We need to specify two boundary conditions for each variable. This is done by the conditions at the wall ðy ¼ 0Þ and in the free stream of the enviornment outside of the boundary layer ðy ¼ Þ. Usually the environment conditions are known. At y ¼ , T ¼ T1 YO2 ¼ YO2 ;1 ð0:233 for airÞ YP ¼ YP;1 ð 0 for airÞ YF ¼ 0 ðif no fuel is leftÞ

ð9:25Þ

T ¼ Tv ; the vaporization temperature YF ¼ YF;o within the condensed phase; i:e: y ¼ 0

ð9:26aÞ

At y ¼ 0,

238

BURNING RATE

Surface boundary conditions (a) and (b)

Figure 9.8

However, we wish to express YF for the gas phase y ¼ 0þ . This requires a more careful analysis for a control volume bounding the surface (Figure 9.8). From Figure 9.8(a), the mass flow rate of the liquid-condensed phase fuel must balance out the gas phase bulk transport at the flow rate m_ 00F and the diffusion flow rate due to a concentration gradient in the gas phase. Note that there is no concentration gradient in the liquid since it is of a uniform concentration. If we have a pure condensed phase fuel, YF;o ¼ 1; in general, the surface concentration of fuel gives Across

y¼0:

m_ 00F YF;o

¼

m_ 00F YF ð0Þ

dYF

D dy y¼0

ð9:26bÞ

Similarly, we can write for no P and O2 absorbed in the condensed phase, y ¼ 0; y ¼ 0;

dYO2 dy dY P 0 ¼ m_ 00F YP D dy

0 ¼ m_ 00F YO2 D

ð9:26cÞ ð9:26dÞ

The corresponding boundary condition for energy is essentially Equation (9.8a), which follows from Figure 9.8(b), i.e. y ¼ 0;

0 ¼ m_ 00F L k

dT dy

ð9:26eÞ

A certain similarity marks these equations and boundary conditions for the T and Yi values. This recognition compels us to seek some variable transformations that will bring simplicity to the equations, and ultimately their solution. We seek to normalize and to simplify. First we recognize that for air and for diffusion in air D 10 5 m2 =s

DIFFUSIVE BURNING OF LIQUID FUELS

239

or Pr ¼

1

and Schmidt number, Sc =D 1, and therefore Lewis number;

Le

Sc ¼ 1 Pr D

ð9:27Þ

This is called the Le ¼ 1 assumption, and is a reasonable approximation for combustion in air. Consequently, we define D ¼ k=cp . Our transformation will take several steps: 1. Let T T T1 ;

i Yi Yi;1

ð9:28Þ

2. Subtract one equation from another among Equations (9.23) and (9.24). For example, for T and i ¼ F, m_ 00F

d T d2 T hc

2 ¼ m_ F 000 dy dy cp

ð9:29aÞ

Operating by hc d2 F 00 d F 000

2 ¼ m_ F m_ F cp dy dy

ð9:29bÞ

gives m_ F

d FT d2 FT

¼0 dy dy2

ð9:30Þ

where

FT T þ

hc F cp

or

FT ¼ ðT T1 Þ þ

hc YF cp

ð9:31Þ

(recall that YF;1 ¼ 0). The -equation is now homogeneous (RHS ¼ 0) with boundary conditions, y ¼ ; y ¼ 0;

¼ F ¼ 0 or FT ¼ 0 T m_ 00 L 0¼ F cp dy

ð9:32aÞ

240

BURNING RATE

and m_ 00F YF;o ¼ m_ 00F YF ð0Þ

d F dy

In the format at y ¼ 0, 0 ¼ m_ 00F

L hc d FT þ ðYF ð0Þ YF;o Þ cp cp dy

ð9:32bÞ

Since the term in brackets on the RHS is constant, we can conveniently normalize FT as bFT

FT L=cp þ ðhc =cp ÞðYF ð0Þ YF;o Þ

or bFT ¼

cp ðT T1 Þ þ hc YF L hc ðYF;o YF ð0ÞÞ

(9.33)

Recall YF;o ¼ 1 for a pure fuel condensed phase, and YF ð0Þ is not known. For a liquid fuel, YF ð0Þ is found from the Clausius–Clapeyron equation provided we know Tð0Þ. In the b format, all of the equations – three from appropriate algebraic additions – are of the form m_ 00F

db d2 b

2 ¼0 dy dy

ð9:34Þ

with y ¼ ;

b¼0

and y ¼ 0;

0 ¼ m_ 00F

db dy

It can be shown that the other b values are bOT ¼

cp ðT T1 Þ þ ðYO2 YO2 ;1 Þðhc =rÞ L

(9.35)

YF ðYO2 YO2 ;1 Þ=r YF ð0Þ YF;o

(9.36)

bOF ¼

DIFFUSIVE BURNING OF LIQUID FUELS

241

which follows the same conditions as Equation (9.34). Note that if the more complete boundary layer equations or the full Navier–Stokes equations were considered, Equation (9.34) would be a partial differential equation containing the additional independent variables x, z and time t. However, it could still be put into this more compact form. This variable transformation, which has essentially eliminated the nonlinear chemical source terms involving m_ F 000 , is known as the Shvab–Zeldovich variable transformation (S–Z). Let us integrate Equation (9.34) using the boundary conditions. Integrating once with a ¼ m_ 00F =, ab

db ¼ ac1 dy

Letting ¼ b c1 , so that Z

d ¼

Z

y

ady 0

gives ¼ ð0Þeay from y ¼ ;

0 c1 ¼ ð0Þea

y ¼ 0;

0 ¼ a ð0Þ a e0

and

Therefore, ð0Þ ¼ 1 and c1 ¼ ea or b ¼ eay ea

(9.37)

is our solution. However, we are far from a meaningful answer. Our objective is to find the ‘burning rate’, m_ 00F , and its dependencies.

9.2.3

Burning rate – an eigenvalue

There is only one value of m_ 00F – an eigenvalue – that will satisfy a given set of environmental and fuel property conditions. How can this be found? We must be creative. We have already pushed the chemical kinetic terms out of the problem, but surely they must not have disappeared in reality. Let us recall from the discussion of liquid evaporation that thermodynamically we have a property relationship between fuel vapor concentration and surface temperature, YF ð0Þ ¼ f ðTð0ÞÞ

or

bOT ð0Þ ¼ bOF ð0Þ

242

BURNING RATE

From Equation (9.37), we evaluate b at y ¼ 0 and express m_ 00F

¼

k ln½1 bð0Þ cp

ð9:38Þ

If we know bð0Þ, we have our answer, but we have several alternatives for b: bFT ; bOT or bOF . Any of these is fine. We chose bOT because it gives us the most expedient approximate answer: bOT ð0Þ ¼

cp ðTð0Þ T1 Þ þ ðYO2 ð0Þ YO2 ;1 Þðhc =rÞ L

ð9:39Þ

We do expect YO2 ð0Þ to be zero at the surface for combustion within the boundary layer since the flame reaction is fast and no oxygen is left. This must be clearly true even if the chemistry is not so fast. Moreover, since we are heating the surface with a nearby flame that approaches an adiabatic flame temperature, we would expect a high surface temperature. For a liquid fuel, we must have Tð0Þ Tboil

at

p1

and we use the equality to the boiling point as an approximation. For a solid, we expect Tð0Þ Tig and again might use the equality or some slightly higher temperature. In general, Tð0Þ ¼ Tv

ð9:40Þ

a specified vaporization temperature. This choice is empirical, unless we have steady burning of pure liquid, where the boiling point is valid, although we have not proven it. To prove it, we could equate bOT ð0Þ and bOF ð0Þ using the thermodynamic equilibrium to obtain YF ðTÞ, the saturation condition. Proceeding, we obtain from Equation (9.38), m_ 00F

¼

k lnð1 þ BÞ cp

(9.41)

The dimensionless group in the log term is called the Spalding B number, named after Professor Brian D. Spalding who demonstrated its early use [5]. From Equation (9.39), the surface conditions give B

YO2 ;1 ðhc =rÞ cp ðTv T1 Þ L

(9.42)

This group expresses the ratio of chemical energy released to energy required to vaporize the fuel (per unit mass of fuel). It is clear it must be greater than 1 to achieve sustained

DIFFUSION FLAME VARIABLES

243

burning. B contains both environmental and fuel property factors. Note a subtle point that cp in the numerator is for the gas phase while in L, where L hfg þ cp;o ðTv Ti Þ cp;o being the specific heat of the condensed phase.

9.3 Diffusion Flame Variables 9.3.1

Concentrations and mixture fractions

Equation (9.41) constitutes a fundamental solution for purely convective mass burning flux in a stagnant layer. Sorting through the S–Z transformation will allow us to obtain specific stagnant layer solutions for T and Yi . However, the introduction of a new variable – the mixture fraction – will allow us to express these profiles in mixture fraction space where they are universal. They only require a spatial and temporal determination of the mixture fraction f . The mixture fraction is defined as the mass fraction of original fuel atoms. It is as if the fuel atoms are all painted red in their evolved state, and as they are transported and chemically recombined, we track their mass relative to the gas phase mixture mass. Since these fuel atoms cannot be destroyed, the governing equation for their mass conservation must be m_ 00F

df d2 f

D 2 ¼ 0 dy dy

ð9:43Þ

since these atoms have bulk transport and diffusion of the species containing the fuel atoms. Although the mixture fraction has physical meaning and can be chemically measured, it can also be thought of as a fictitious property that satisfies a conservation of mass equation as an inert species composed of the original fuel atoms. The boundary conditions must be At

y¼:

f ¼ 0 ðno fuel in the free streamÞ

ð9:44aÞ

At

y¼0:

f ¼ f ð0Þ

ð9:44bÞ

Also from Equation (9.26b), the transport of fuel at y ¼ 0 is m_ 00F YF;o ¼ m_ 00F f ð0Þ D

df dy

ð9:44cÞ

If we define bf

f f ð0Þ YF;o

(9.45)

244

BURNING RATE

then it can be verified that bf perfectly satisfies Equation (9.34) with the identical boundary conditions as do all of the b values. Thus, 00

00

bFT ¼ bOT ¼ bOF ¼ bPF ¼ ¼ bf ¼ ecp m_ F y=k ecp m_ F =k

ð9:46Þ

Before bringing this analysis to a close we must take care of some unfinished business. The chemical kinetic terms, i.e. m_ 000 F , have been transformed away. How can they be dealt with? Recall that we are assuming tchem tDiff (or tmix, if turbulent flow). Anyone who has carefully observed a laminar diffusion flame – preferably one with little soot, e.g. burning a small amount of alcohol, say, in a whiskey glass of Sambucca – can perceive of a thin flame (sheet) of blue incandescence from CH radicals or some yellow from heated soot in the reaction zone. As in the premixed flame (laminar deflagration), this flame is of the order of 1 mm in thickness. A quenched candle flame produced by the insertion of a metal screen would also reveal this thin yellow (soot) luminous cup-shaped sheet of flame. Although wind or turbulence would distort and convolute this flame sheet, locally its structure would be preserved provided that tchem tmix . As a consequence of the fast chemical kinetics time, we can idealize the flame sheet as an infinitessimal sheet. The reaction then occurs at y ¼ yf in our one dimensional model. The location of yf is determined by the natural conditions requiring At

YF ¼ YO2 ¼ 0

y ¼ yf :

ð9:47Þ

This condition defines the thin flame approximation, and implies that the flame settles on a location to accommodate the supply of both fuel and oxygen. This condition also provides a mathematical closure for the problem since all other solutions now follow. We will illustrate these. Let us consider the solutions for YF and YO2 in the mixture fraction space. From Equations (9.45) and (9.46), for any b, b¼

f f ð0Þ YF;o

and bð0Þ ¼

f ð0Þ f ð0Þ YF;o

or f ð0Þ ¼

bð0ÞYF;o bð0Þ 1

and finally b¼

f ðbð0Þ 1Þf ¼ bð0ÞYF;o =½bð0Þ 1 YF;o YF;o

DIFFUSION FLAME VARIABLES

245

or f ¼

YF;o b bð0Þ 1

(9.48)

Now select b ¼ bOF . Therefore, from Equation (9.36),

f ¼Y

YF;o

h

YF ðYO2 YO2 ;1 Þ=r YF ð0Þ YF;o

F ð0Þ ðYO2 ð0Þ YO2 ;1 Þ=r

YF ð0Þ YF;o

i

Y ð0Þ Y

YFF ð0Þ YF;o F;o

and rearranging, taking YO2 ð0Þ ¼ 0, f ¼

YF ðYO2 YO2 ;1 Þ=r YO2 ;1 =ðrYF;o Þ þ 1

ð9:49Þ

Since at the flame sheet YO2 ¼ YF ¼ 0, this is the definition of a stoichiometric system so that f ðyf Þ fst ¼

YO2 ;1 =r YO2 ;1 =ðrYF;o Þ þ 1

(9.50)

By substituting Equation (9.50) into Equation (9.49), it can be shown that YO2 YF;o fst f ¼ YF

þ fst r YF;o

ð9:51Þ

We can now choose to represent YO2 and YF in y or in the f ðyÞ variable as given by Equation (9.46). On the fuel side of the flame, from Eq. (9.51), 0 y yf ; YO2 ¼ 0 f fst YF ¼ 1 fst =YF;o

ð9:52Þ

Similarly, on the oxygen side, yf y , YF ¼ 0 and YO2 ¼

rðf fst Þ YO2 ;1 ðfst f Þ ¼ 1 fst =YF;o fst

Note at y ¼ 0 that YF ð0Þ ¼

f ð0Þ fst 1 fst =YF;o

ð9:53Þ

246

BURNING RATE

Figure 9.9 Oxygen–fuel mass concentrations in mixture fraction space

by Equation (9.45b), and at y ¼ , þrfst YO2 ¼ 1 fst =YF;o YO =½YO2 =rYF;o þ 1 ¼ 2 1 ½YO2 ;1 =ðrYF;o Þ ½YO2 ;1 =ðrYF;o Þ þ 1 YO2 ;1 ¼ YO2 ;1 ¼ YO2 ;1 =ðrYF;o Þ þ 1 YO2 ;1 =ðrYF;o Þ The results are plotted in Figure 9.9. Typical values of fst are, for air ðYO2 ;1 ¼ 0:233Þ and r 1 to 3, roughly 0.05–0.2. The results are linear in f but nonlinear in y. This is true for all of the variables YPi and T. The relationships for Yi and T in f are universal, independent of geometry, flow conditions, etc., provided the thin flame – laminar – assumption holds. Even for unsteady, turbulent flows this conclusion still holds under a local thin flame approximation. Let us recapitulate. We have achieved a solution to boundary-layer-like burning of a steady liquid-like fuel. A thin flame or fast chemistry relative to the mixing of fuel and oxygen is assumed. All effects of radiation have been ignored – a serious omission for flames of any considerable thickness. This radiation issue cannot easily be resolved exactly, but we will return to a way to include its effects approximately.

9.3.2

Flame temperature and location

The position of the flame sheet, yf , can be determined in a straightforward manner. From Equations (9.45), (9.46) and (9.50), YO2 ;1 =r e eðyf =Þ ¼ ð9:54Þ ½YO2 ;1 =ðrYF;o Þ þ 1ðYF;o f ð0ÞÞ

DIFFUSION FLAME VARIABLES

247

where cp m_ 00F =k. From Equations (9.39), (9.42) and (9.46), bf ð0Þ ¼ B, the Spalding number. Therefore, it can be shown from Equation (9.45) that YF;o f ð0Þ ¼

YF;o 1þB

Realizing that e ¼ 1 þ B from Equation (9.41), it follows that eyf = ¼

1þB YO2 ;1 =ðrYF;o Þ þ 1

or yf ln½ð1 þ BÞ=ðYO2 ;1 =ðrYF;o Þ þ 1Þ ¼ lnð1 þ BÞ

(9.55)

The flame temperature is found from using Equations (9.35), (9.48) and (9.50), evaluated at y ¼ yf or f ¼ fst, YF;o ½cp ðTF T1 Þ YO2 ;1 hc =r L ¼ YO2 ;1 c ðT

T Þ

Y p v 1 O2 ;1 ðhc =rÞ rYF;o þ 1

1 L YO2 ;1 =r

After rearranging, we find the flame enthalpy rise as cp ðTf T1 Þ ¼

YF;o hc L þ cp ðTv T1 Þ 1 þ rYF;o =YO2 ;1

(9.56a)

Alternatively, this can be expressed as cp ðTf T1 Þ B o ðro 1Þ ro ¼ L 1 þ ro YO2 ;1 ro ¼ rYF;o cp ðTv T1 Þ o ¼ L

ð9:56bÞ

The quantity rYF;o =YO2 ;1 is the ratio of stoichiometric oxygen to fuel mass ratio divided by the available or supplied oxygen to fuel mass fractions. In contrast, for the premixed adiabatic flame temperature, cp ðTf;ad T1 Þ ¼ YF;u hc

ð9:57Þ

the analog of Equation (4.54). As in estimating the lower limit concentration that permits a premixed flame to exist by requiring Tf;ad ¼ 1300 C, a similar requirement could be imposed to define conditions for maintaining a diffusion flame. We will use this to estimate extinction conditions.

248

BURNING RATE

9.4 Convective Burning for Specific Flow Conditions The stagnant layer analysis offers a pedagogical framework for presenting the essence of diffusive burning. For the most part the one-dimensional stagnant layer approximated a two-dimensional boundary layer in which ¼ ðxÞ, with x the flow direction. For a convective boundary layer, the heat transfer coefficient, hc , is defined as hc ðTR Ts Þ ¼ k

dT dy y¼0

ð9:58Þ

where TR is a reference temperature and Ts is the wall temperature. For a flame this might be approximated with TR ¼ Tf , as the flame temperature, hc ðTf Ts Þ

kðTf Ts Þ

for the stagnant layer, or hc

k

ð9:59Þ

This crude approximation allows us to extend the stagnant layer solution to a host of convective heat transfer counterpart burning problems. Recall that for Equations (9.41) and (9.42), we can write m_ 00F

hc lnð1 þ BÞ cp

(9.60)

We wish to write this as m_ 00F

hc lnð1 þ BÞ ¼ B B cp

ð9:61Þ

where the significance of the blocking factor, lnð1 þ BÞ=B will become apparent. Substituting for B and recalling the heat flux from Equation (9.6), Equation (9.61) becomes q_ 00 ¼ hc

lnð1 þ BÞ B

YO2 ;1 hc

ðTv T1 Þ cp r

ð9:62Þ

Thus we see that the reference temperature should have been YO2 ;1 hc =ðcp rÞ þ T1 , which is slightly different from Equation (9.56a). Nevertheless, we see that lnð1 þ BÞ q_ ¼ hc ðTR Tv Þ B 00

CONVECTIVE BURNING FOR SPECIFIC FLOW CONDITIONS

249

The blocking factor can alternatively be written from Equation (9.60) as lnð1 þ BÞ cp m_ 00F =hc ¼ B expðcp m_ 00F =hc Þ 1

ð9:63Þ

which approaches 1 as m_ 00F ! 0. Hence the heat transfer coefficient in the presence of mass transfer is smaller than hc , the heat transfer coefficient without mass transfer. The blocking factor acts to enlarge the boundary later due to blowing caused by the vaporized fuel. This transverse flow makes it more difficult to transfer heat across the boundary layer. Moreover, in Equation (9.61), it is seen that for mass transfer in combustion, the mass transfer coefficient is hm

hc lnð1 þ BÞ B cp

ð9:64Þ

and the driving force for this transfer is B. B acts like T for heat, Y for mass or voltage for electrical current flow. To obtain approximate solutions for convection heating problems, we only need to identify a heat transfer problem that has a given theoretical or empirical correlation for hc . This is usually given in the form of the Nusselt number (Nu), hc x Nux ¼ f ðRex1 ; Grx2 PrÞ k

ð9:65Þ

The corresponding dimensionless burning rate (called the Sherwood number ðShÞ in pure mass transfer) should then give a functional relationship of the form m_ 00F cp x ¼ f ðNux ; BÞ k

ð9:66Þ

or more generally may contain other factors that we have seen, e.g. o ¼ cp ðTv T1 Þ=L, ro ¼ YO2 ;1 =ðrYF;o Þ, etc. Figure 9.10 gives some pure convective heat transfer formulas and their associated geometries. Since the heat transfer formula will contain T=T1 in the Grashof number, i.e. Grx ¼

gTx3 T1 2

it is suggested to take T=T1 3 since it could be argued that values of 1 to 7 should be used for this type of approximation. Remember, this is only an approximate procedure. The effect of turbulence is not just in hc , but is also in r. In turbulent flow, due to unmixedness of the fuel and oxidizer, r can be replaced by nr, where n is the excess oxidizer fraction. There have been a number of direct solutions in the form of Equation (9.65) for convective combustion. These have been theoretical – exact or integral approximations to the boundary layer equations, or empirical – based on correlations to experimental data. Some examples are listed below:

250

BURNING RATE

Figure 9.10

Pure convective heat transfer [6]

1. Forced convection, burning of a flat plate. This classical exact analysis following the well-known Blasius solution for incompressible flow was done by Emmons in 1956 [7]. It includes both variable density and viscosity. Glassman [8] presents a functional fit to the Emmons solution as u x1=2 lnð1 þ BÞ m_ 00F cp x 1 ¼ 0:385 Pr k B0:15

ð9:67Þ

2. The corresponding laminar natural convection burning rate on a vertical surface was done by Kosdon, Williams and Buman [9] and by Kim, deRis and Kroesser [10]. The latter gives m_ 00F cp x ¼ 3 Prw Grw1=4 FðB; o ; ro Þ kw

ð9:68Þ

where the transport properties are evaluated at Tw and F is given in Figure 9.11(a).

Figure 9.11

(a) Laminar burning rate of a vertical plate under natural convection by Kim, deRis and Kroesser [10]. (b) Turbulent burning rate of a vertical plate under natural connection by Ahmad and Faeth [11]

252

BURNING RATE

3. Turbulent convective burning of vertical plate. An approximate solution matched to data was given by Ahmad and Faeth [11]. The average burning rate m_ 00F for a distance x measured from the start of the plate is given by the formula below: m_ 00F xcp 0:0285Grx0:4 Pr 0:73 ¼ k

ð9:69Þ

where g cos x3 ð measured from the verticalÞ 2 1=2 1þB 1 þ 0:5 Pr=ð1 þ BÞ 1=4 ¼ B lnð1 þ BÞ 3ðB þ o Þf þ o cp ðTv T1 Þ o ¼ L 1=3 f ¼ 1 FOf

Gr ¼

L 4cp T1

ro ðB þ 1Þ Bðro þ 1Þ YO2 ;1 ro ¼ rYF;o Z x 1 m_ 00F ¼ m_ 00 dx x 0 F

FOf ¼

This solution is a bit tedious, but complete and consistent with data on liquid fuels saturated into inert plates, as displayed in Figure 9.11(b). Table 9.2 gives some typical estimated or exact properties in order to compute the B numbers. It is seen in some cases, particularly for charring materials, that the B numbers for burning in air are less than 1. This implies that purely convective burning is not possible. Indeed, we have ignored surface radiation loss, which can be relatively considerable (at 350 C, 8.5 kW/m2). The only hope of achieving ‘steady’ burning is to provide additional heat flux if burning is not possible on its own. Therefore, not only is there a minimum heat flux to achieve ignition, but there is also a minimum external heat flux needed to maintain burning. Example 9.1 Methanol is saturated into a thick board of porous ceramic fibers. The board is maintained wet with a supply temperature at ambient of 20 C. A steady air flow is directed across the horizontal board at 3 m/s. The board is 10 cm in the flow directions, and is placed flush with the floor. Use the following property data: For air: ¼ 15 10 6 m2 =s k ¼ 25 10 3 W=m K Pr ¼ 0:7 ¼ 1:1 kg=m3 cp ¼ 1:05 kJ=kg K

CONVECTIVE BURNING FOR SPECIFIC FLOW CONDITIONS

253

Table 9.2 Estimated fuel properties for B-number estimates (from various sources) Material

hac (kJ/g)

L (kJ/g)

Tvb ( C)

B-estimatec

69 98 125 80 110 218 64 78 117 56

6.7 6.2 5.7 6.2 5.9 5.2 2.5 3.1 3.6 5.2

360 330 500 300 350

0.75 0.89 0.68 1.4 0.91

300 380 380 300 350

0.56 0.22 0.29 0.30 0.58

Liquids n-Hexane n-Heptane n-Octane Benzene Toulene Naphthalene Methanol Ethanol n-Butanol Acetone

42 41 41 28 28 30 19 26 35 28

0.45 0.48 0.52 0.48 0.50 0.55 1.2 0.97 0.82 0.58

Solids 38 3.6 38 3.1 27 3.8 24 2.0 27 3.0 Solids, charring Polyurethane foam, rigid 17 5.0 Douglas fir 13 12.5 Redwood 12 9.4 Red oak 12 9.4 Maple 13 4.7 Polyethylene Polypropylene Nylon Polymethylmethacrylate Polystyrene

a

Based on flaming conditions and the actual heat of combustion. Boiling temperature for liquids; estimated ignition temperature for solids. c YF;o ¼ 1; YO2 ;1 ¼ 0:233; T1 ¼ Ti ¼ 25 C, cp ¼ 1 kJ=kg K. b

Methanol boiling point ¼ 337 K ¼ 64 C Heat of vaporization, hfg ¼ 1:1 kJ=g Liquid specific heat ¼ 2.5 kJ/kg K Assume steady burning and ignore radiation effects. (a) Assess that the flow is laminar. (b) Calculate the B number. (c) What is the steady burning rate g/s if the plate is 20 cm wide? Use the Emmons result given by Equation (9.67). (d) Calculate the average net heat flux to the surface. (e) What is the surface radiation heat flux emitted assuming a blackbody? (f) What is the role of the ceramic board in ignition and steady burning? Solution (a) Rex ¼

u1 x ð3 m=sÞð0:1 mÞ ¼ ¼ 2 104 15 10 6 m2 =s

254

BURNING RATE

Flow is laminar since Re < 5 105 , transition. (b) B ¼

YO2 ;1 hc =r cp ðTv T1 Þ L

From Table 2.3, hc ¼ 20:0 kJ=g From the stoichiometric equation for complete combustion, CH4 O þ 1:5 O2 ! CO2 þ 2 H2 O ð1:5Þð32Þ ¼ 1:5 g O2 =g fuel 32 YO2 ;1 hc =r ¼ ð0:233Þð20:0 kJ=gÞ=1:5 ¼ 3:11 kJ=g r¼

cp ðTv T1 Þ ¼ ð1:05 kJ=kg K 10 3 kg=gÞð64--20Þ ¼ 0:046 kJ=g L ¼ hfg þ cl ðTv T1 Þ ¼ 1:10 kJ=g þ ð2:5 10 3 kJ=g KÞð64--20Þ ¼ 1:21 kJ=g B ¼ ð3:11 0:046Þ=1:21 ¼ 2:3 (c) From Equation (9.67), using air properties for the gas, 1=2 25 10 3 W=m K 3 m=s m_ 00F ðxÞ ¼ ð0:7Þ ð0:385Þ 1:05 J=g K 15 10 6 m2 =s lnð1 þ 2:53Þ 1=2 x ; x in m ð2:53Þ0:15 pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ ¼ 1:097= xðmÞ in g=m2 s The burning rate over x ¼ 0:10 m of width 0.20 m, Z 0:10 m 1:097 m_ ¼ ð0:20 mÞ dx x1=2 0 m_ ¼ 0:139 g=s (d) From Equation (9.6), q_ 00 ¼ m_ 00 L 0:139 g=s ð1:21 kJ=gÞ ¼ 0:20 0:10 m2 ¼ 8:4 kW=m2 4 Þ; (e) q_ 00r ¼ ðTv4 T1

¼ 5:67 10

11

assuming ambient surroundings kW=m2 K4 ½ð337Þ4 ð293Þ4

¼ 0:313 kW=m2 Hence, re-radiation is negligible.

RADIATION EFFECTS ON BURNING

255

(f) In steady burning, the ceramic board functions as a wick and achieves the temperature of the methanol liquid in depth. For the ignition process, the ceramic board functions as a heat sink along with the absorbed methanol. The surface temperature must achieve the flash (or fire) point of the methanol to ignite with a pilot. The ignition process cannot be ignored in the unsteady burning problem since it forms the initial condition.

9.5 Radiation Effects on Burning The inclusion of radiative heat transfer effects can be accommodated by the stagnant layer model. However, this can only be done if a priori we can prescribe or calculate these effects. The complications of radiative heat transfer in flames is illustrated in Figure 9.12. This illustration is only schematic and does not represent the spectral and continuum effects fully. A more complete overview on radiative heat transfer in flame can be found in Tien, Lee and Stretton [12]. In Figure 9.12, the heat fluxes are presented as incident (to a sensor at T1 ) and absorbed (at Tv ) at the surface. Any attempt to discriminate further for the radiant heating would prove tedious and pedantic. It should be clear from heat transfer principles that we have effects of surface and gas phase radiative emittance, reflectance, absorptance and transmittance. These are complicated by the spectral character of the radiation, the soot and combustion product temperature and concentration distributions, and the decomposition of the surface. Reasonable approximations that serve to simplify are: 1. Surfaces are opaque and approach emittance and absorptance of unity. 2. External radiant flux is not affected by the flame, e.g. thin flame. 3. Flame incident radiant flux ðq_ 00f;r Þ and total radiant energy ðXr Q_ Þ might be estimated from measurements. 4. External radiant heat flux ðq_ 00e Þ is expressed as that measured by a heat flux meter maintained at T1 . Correspondingly, the surface loss is given with respect to T1 as 4 ðTv4 T1 Þ. With the heat flux specifications in Figure 9.12, the energy boundary condition at y ¼ 0 becomes, for steady burning (see Equations (9.6) and (9.8)), dT 00 4 m_ F L ¼ k þq_ 00 þ q_ 00e ðTv4 T1 Þ ð9:70Þ dy y¼0 f;r

Figure 9.12

Radiative heat transfer in burning

256

BURNING RATE

If we can define the radiative terms a priori, we are simply adding a constant to the RHS. An expedient transformation reduces the boundary condition to its original convective form: m_ 00F Lm ¼ k

dT dy

at

y¼0

ð9:71Þ

where Lm L

4 Þ q_ 00f;r þ q_ 00e ðTv4 T1 00 m_ F

(9.72)

is defined as an equivalent, modified ‘L’. The net addition of radiant heat makes the burning rate increase as one would expect. It might also make it burn where a material will not normally. This is an important point because many common materials and products will not burn without additional heat flux. While this addition may seem artificial, it is the key to disastrous fire growth since special configurations favor heat feedback ðq_ 00e Þ that can promote fire growth. Limited testing of a material by applying a small flame might cause ignition, but not necessarily sustained burning or spread; this can be misleading to the evaluation of a material’s general performance in fire. The introduction of Equation (9.71) for Equation (9.26e) makes this a new problem identical to what was done for the pure diffusion/convective modeling of the burning rate. Hence L is simply replaced by Lm to obtain the solution with radiative effects. Some rearranging of the stagnant layer case can be very illustrative. From Equations (9.61) and (9.42) we can write m_ 00F ¼

hc lnð1 þ BÞ YO2 ;1 hc =r cp ðTv T1 Þ 4 Þ=m _ 00F B L ½q_ 00f;r þ q_ 00e ðTv4 T1 cp

or

m_ 00F L ¼

hc YO2 ;1 hc ð1 Xr Þ cp ðTv T1 Þ cp e 1 r 4 þ q_ 00f;r þ q_ 00e ðTv4 T1 Þ

ð9:73Þ

with ¼ cp m_ 00F =hc and where we have added ð1 Xr Þ to account for radiated energy ðXr Q_ Þ eliminated from the flame. Although this requires an iterative solution for m_ 00F, its form clearly displays the heat flux contributions to vaporization. The LHS is the energy required to vaporize under steady burning and the RHS is the net absorbed surface heat flux, the first term being the convective heat flux. Typical convection ranges from 5 to 10 kW/m2 for turbulent buoyant flows but as high as 50–70 kW/m2 for laminar natural convection. For a forced convection laminar flame, such as a small jet flame, the convective heat flux can be 100–300 kW/m2. Therefore, convective effects cannot generally be dismissed although they are a relatively small component in turbulent natural fires. Figure 9.13 displays the separation of average flame convective and radiative components for burning pools of PMMA at diameters ranging from 0.15 to 1.2 m.

RADIATION EFFECTS ON BURNING

Figure 9.13

257

Average flame heat flux in PMMA pool fires (from Modak and Croce [13] and Iqbal and Quintiere [14])

Radiation increases with scale while convection drops as the diameter increases. Local resolution of these flame heat fluxes were estimated by Orloff, Modak and Alpert [15] for a burning 4 m vertical wall of PMMA. It is clearly seen that flame radiative effects are significant relative to convection. In Figure 9.14 we see a more classical demonstration of the range of burning rate behavior of a pool fire. Below a diameter of 25 cm the burning rate is laminar with hc / D 1=4 ; afterwards it is turbulent as hc / D0 or D1=5 at most. However, with

Figure 9.14

Local flame heat flux to a burning PMMA vertical surface (from Orloff, Modak and Alpert [15])

258

BURNING RATE

Figure 9.15

Regimes of the steady burning rate for methanol: D < 25 cm laminar, D > 25 cm turbulent, D > 100 cm radiation saturated [16, 17]

increasing diameter, the primary cause of the increase in the burning rate is due to flame radiation. This follows as the flame emittance behaves as f 1 e D which approaches 1 at about D 100 cm. Since L ¼ 1:21 kJ=g for methanol, we see from Figure 9.15 that (1) the maximum net surface heat flux at about D ¼ 5 cm is q_ 00 ¼ ð27 g=m2 sÞð1:21 kJ=gÞ ¼ 32:7 kW=m2 (2) a minimum at the onset of turbulent flow of q_ 00 ¼ ð12 g=m2 sÞð1:21 kJ=gÞ ¼ 14:5 kW=m2 and (3) a maximum under ‘saturated’ flame radiation conditions of q_ 00 ¼ ð21 g=m2 sÞð1:21 kJ=gÞ ¼ 25:4 kW=m2 Note that the re-radiation loss from the methanol surface at Tb ¼ 64 C is only about 0:3 kW=m2 .

PROPERTY VALUES FOR BURNING RATE CALCULATIONS Table 9.3

259

Asymptotic burning rates (from various sources) g=m2 s

Polyvinyl chloride (granular) Methanol Flexible polyurethane (foams) Polymethymethacrylate Polystyrene (granular) Acetone Gasolene JP-4 Heptane Hexane Butane Benzene Liquid natural gas Liquid propane

16 21 21–27 28 38 40 48–62 52–70 66 70–80 80 98 80–100 100–130

The asymptotic burning rate behavior under saturated flame radiation conditions is a useful fact. It provides an upper limit, at pool-like fuel configurations of typically 1–2 m diameter, for the burning flux. Experimental values exist in the burning literature for liquids as well as solids. They should be thoughtfully used for design and analysis purposes. Some maximum values are listed in Table 9.3. Equation (9.73) can be used to explain the burning behavior with respect to the roles of flame radiation and convection. It can also explain the effects of oxygen and the addition of external radiant heat flux. These effects are vividly shown by the correlation offered from data of Tewarson and Pion [18] of irradiated horizontal small square sheets of burning PMMA in flows of varying oxygen mole fractions. The set of results for steady burning are described by a linear correlation in q_ 00e and XO2 for L ¼ 1:62 kJ=g in Figure 9.16. This follows from Equation (9.73): c _ 00e hc e 1 h cp r ð1 Xr ÞðYO2 YO2 ;1 Þ þ q 00 00 m_ F m_ F;1 ¼ L where m_ 00F;1 is the burning rate under ambient air conditions with YO2 ;1 ¼ 0:233 and q_ 00e ¼ 0. This result assumes that the mass transfer coefficient maintains a fairly constant value over this range of m_ 00F , and that q_ 00f is constant for all conditions of oxygen and radiation. A valid argument can be made for the constancy of q_ 00f as long as the flame is tall (Lf > 2D at least) and does not change color (the soot character is consistent). A tall gray cylindrical flame will achieve a constant emittance for Lf > 2D with respect to its base.

9.6 Property Values for Burning Rate Calculations The polymer PMMA has been used to describe many aspects of steady burning rate theory. This is because it behaves so ideally. PMMA decomposes to its base monomer; although it melts, its transition to the vaporized monomer is smooth; its decomposition

260

BURNING RATE

Figure 9.16

Pool fire burning rates for PMMA as a function of oxygen and external radiation [18]

kinetics give a reasonably constant surface temperature during burning; and steady burning is easy to achieve. However, all brands of PMMA are not identical and values of L may range from 1.6 to 2.8 kJ/g. As materials become more complex in their decomposition, the idealizations made in our theory cannot be fulfilled. Specifically, for charring materials, steady burning is not physically possible. Yet, provided we define consistent ‘properties’ for the steady burning theory, it can yield rational time-average valid burning rates for even charring materials. The approximate nature of this steady modeling approach must always be kept in mind and used appropriately. We have already mentioned that practical data under fire conditions are commonly presented in terms of mass loss of the fuel package (i.e. Equations (9.3), (9.4) and (9.6)). Effectively, this means for fuels in natural fire scenarios, such as furnishings, composites, plastic commodities, etc., we must interpret the steady burning theory in the following manner: 1. hc and L are both based on mass loss, i.e. kJ/g mass loss. 2. YF;o is taken as unity. 3. r is evaluated by assuming that hc =r 13 kJ=g O2 used. 4. Xr , the radiative fraction, depends on flame size and may increase and then decrease with fire diameter. It can range from 0.10 to 0.60.

SUPPRESSION AND EXTINCTION OF BURNING

261

5. Fluid properties can be taken as that of air with or without attention to temperature effects. Because of the large temperature range found in combustion, true property effects with temperature are rarely perfectly taken into account. However, in keeping with standard heat transfer practices of adopting properties at a film temperature, the following are recommended: Air at 1000 C: ¼ 0:26 kg=m3 cp ¼ 1:2 kJ=kg K k ¼ 0:081 W=m K ¼ 1:8 10 4 m2 =s Pr ¼ 0:70

9.7 Suppression and Extinction of Burning Equation (9.73) offers a basis for addressing the effects of variables on a reduction in the burning rate. We have already seen in Figure 9.16 that the reduction of oxygen can reduce the burning rate and lead to an extinction value, namely about 5 g/m2 s for PMMA. Moreover, the correlation shows that if the critical mass flux is indeed constant then there is a relationship between oxygen concentration ðXO2 Þ and external heat flux ðq_ 00e Þ at extinction. The higher q_ 00e , the lower XO2 can be before extinction occurs. For q_ 00e ¼ 0, the extinction value for XO2 by the correlation in Figure 9.16 is 0.16 or 16 %. Note that the ambient temperature T1 was about 25 C in these experiments.

9.7.1

Chemical and physical factors

We will try to generalize these effects for suppression and will adopt a temperature criterion for the extinction of a diffusion flame. Clearly at extinction, chemical kinetic effects become important and the reaction ‘quenches’. The heat losses for the specific chemical dynamics of the reaction become too great. This can be qualitatively explained in terms of Equation (9.12): tchem ðRe2chem Pr 2 Þ

RT tDiff where the dimensionless parameters are as follows: 1. Chemical Reynold’s number, u1 chem ﬃ rﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ kT ¼ Ahc

Rechem ¼ chem

ð9:74Þ

262

BURNING RATE

where chem is a reaction length scale, not unlike the reaction width of a premixed flame. 2. Dimensionless reaction rate parameter RT ¼

E E=ðRTÞ e RT

3. Prandtl number, Pr ¼ =. Extinction becomes likely as tchem =tDiff becomes large, or RT is small or Rechem is large. Provided u1 is not as high and A (the pre-exponential coefficient, dependent on YF;o and YO2 ;1 ) is in a normal range for nonretarded hydrocarbon-based fuels, the principal factor controlling the relative times is RT . The empirical rule for the limiting conditions to achieve the propagation of premixed flames is for T 1300 C at least. At lower temperatures, tchem becomes too great with respect to tDiff ( or tmix ) and the reaction rate is too slow. This results because local heat loss dominates for the slow reaction and slows it further as T drops, causing a cascading downward spiral toward extinction. On the other hand, a high value of u1 will achieve the same result. This can be referred to as a ‘flame stretch’ effect, usually expressed as a velocity gradient, e.g. u1 = for a boundary layer. Blowing out a flame is representative of this effect – it also serves to separate the fuel from the oxygen by this velocity gradient across the flame sheet. For natural fire conditions flame stretch is relatively small compared with commercial combustors. However, flame stretch can still be a factor in turbulent fires, serving to locally extinguish the convoluted flame sheet in high shear regions. However, the fluctuating character of species and temperature fields in turbulent flames play a more complex role than described by flame stretch alone. In addition, the effects of gas phase retardants can change both A and E. If E is increased, our critical temperature criterion for extinction must accordingly be increased to maintain effectively a critical constancy for E=T. These chemical effects are complex and specific, and we will not be able to adequately quantify them. It is sufficient to remember that both velocity (flame stretch) and chemistry (retardant kinetics) can affect extinction. We will only examine the temperature extinction criterion.

9.7.2

Suppression by water and diluents

Since water is used as a common extinguishing agent, it will be included explicitly in our analysis of suppression. Other physical (nonchemically acting) agent effects could also be included to any degree one wishes. We shall only address water. We assume water acts in two ways: 1. Droplets are evaporated in the flame. 2. Droplets are evaporated on the surface. The addition of water vapor is not addressed. We would have to specifically consider the species conservation equations for at least the products CO2 and H2 O. Only water that is

SUPPRESSION AND EXTINCTION OF BURNING

263

evaporated is considered, realizing that much water can be put on a fire but only a fraction finds its mark. To consider the flame evaporation effect (1), the energy equation (Equation (9.23)) must contain the energy loss from the gas phase due to water droplet evaporation. The net energy release rate in the gas phase is 000 000 _ w Lw _ 000 Q_ net ¼ m_ 000 F hc Xr m F hc m

ð9:75Þ

where we have accounted for radiation loss by the flame radiation fraction Xr and the water energy loss by m_ 000 w , the rate of water evaporated per unit volume in the flame, and Lw , the heat of gasification for water. We take Lw ¼ hfg þ cp ðTb T1 Þ ¼ 2:6 kJ=g for water. It is clear that the determination of m_ 000 w is not easy, but well-controlled experiments or CFD (computer fluid dynamic) models have the potential to develop this information. For now we simply represent the water heat loss in the flame as a fraction ðXw;f Þ of the flame energy, _ 000 Q_ 000 net ¼ ð1 Xr Xw;f Þm F hc

ð9:76Þ

From the surface evaporation effect (2), we reformulate the surface energy balance (Equation (9.26e)) as the heat transferred (net) must be used to evaporate the fuel and the water at the surface, y ¼ 0;

k

dT ¼ m_ 00F L þ m_ 00w Lw dy

ð9:77Þ

In the generalization leading to Equation (9.73) we can directly write m_ 00F

hc ¼ lnð1 þ BÞ cp

where B¼

YO2 ;1 ð1 Xr Xw;f Þhc =r cp ðTv T1 Þ Lm

(9.78)

and Lm ¼ L

4 Þ m_ 00w Lw q_ 00f;r þ q_ 00e ðTv4 T1 00 m_ F

In other words, we replace hc by ð1 Xr Xw;f Þhc and L by Lm everywhere in our stagnant layer solutions. The solution for flame temperature becomes, from Equation (9.56), cp ðTf T1 Þ ¼

YF;o ð1 Xr Xw;f Þhc Lm þ cp ðTv T1 Þ 1 þ rYF;o =YO2 ;1

(9.79)

264

BURNING RATE

These equations give us the means to estimate physical effects by the reduction of oxygen or addition of water on suppression. The equation for flame temperature with a critical value selected as 1300 C gives a basis to establish extinction conditions. Example 9.2 Estimate the critical ambient oxygen mass fraction needed for the extinction of PMMA. No external radiation or water is added and T1 ¼ 25 C. Compute the mass burning flux just at extinction assuming hc ¼ 8 W=m2 K. For PMMA use a value of L ¼ 1:6 kJ=g, hc ¼ 25 kJ=g and Tv ¼ 360 C. Solution At extinction we expect the flame to be small so we will assume no flame radiative heat flux and Xr ¼ 0. This is not a bad approximation since near extinction soot is reduced and a blue flame is common. For PMMA, YF;o ¼ 1. Substituting values into Equation (9.79), we obtain (approximating Lm ¼ L) ð1:2 10 3 kJ=g KÞð1300 25ÞK ð1Þð1Þ25 kJ=g 1:6 kJ=g þ ð1:2 10 3 kJ=g KÞð360 25ÞK 1 þ ð25=13Þð1Þ=YO2 ;1 23:8 1:53 ¼ 1 þ 1:92=YO2 ;1 2:94 1:53 þ ¼ 23:8 YO2 ;1 ¼

YO2 ;1 ¼ 0:132 or XO2 ;1 ¼ ð0:132Þð29=32Þ ¼ 0:12 molar This molar value compares to 0.16 for correlation of the data in Figure 9.16. This follows from the correlation at zero external flux and for the critical mass flux at extinction, 5 g/m2 s. The burning rate is found from our basic general equation (Equation (9.78)), where YO2 ;1 hc =r cp ðTv T1 Þ L ð0:132Þð13Þ ð1:2 10 3 Þð360 25Þ ¼ 1:6 ¼ 0:821

B¼

Then m_ 00F ¼ ¼

hc lnð1 þ BÞ cp ð8 W=m2 KÞ lnð1:82Þ ð1:2 J=g KÞ

¼ 4:0 g=m2 s

at extinction

SUPPRESSION AND EXTINCTION OF BURNING

Figure 9.17

265

Burning rate of vertical PMMA slabs versus external radiant flux for various water application rates [19]

This compares to 5 g/m2 s of Figure 9.16. The effect of a water spray on the burning of a vertical PMMA slab was investigated by Magee and Reitz [19]. Their results show a linear behavior of the burning rate with external radiant heating and the applied water rate per unit area of PMMA surface as displayed in Figure 9.17. Assuming steady burning, which was sought in the experiments, we can apply Equations (9.78) and (9.79). We assume the following: Xr ¼ 0 Xw;f ¼ 0 for simplicity, since we have no information on these quantities. However, for a wall flame, we expect them to be small. By Equation (9.78) we can write m_ 00F L

00 hc YO2 ;1 hc m_ F cp =hc ¼ 00 cp em_ F cp =hc 1 r cp ðTv T1 Þ

4 Þ m_ 00w Lw þq_ 00f;r þ q_ 00e ðTv4 T1

(9.80)

266

BURNING RATE

and from Equation (9.79) by substituting for Lm, rYF;o _m00F cp ðTf T1 Þ 1 þ

cp ðTv T1 Þ YO2 ;1 4 ¼ m_ 00F YF;o hc m_ 00F L þ q_ 00f;r þ q_ 00e ðTv4 T1 Þ m_ 00w Lw

or rYF;o m_ 00F cp ðTf T1 Þ 1 þ

cp ðTv T1 Þ þ L YF;o hc YO2 ;1 4 ¼ q_ 00f;r þ q_ 00e ðTv4 T1 Þ m_ 00w Lw

ð9:81aÞ

With Tf ¼ 1300 C, T1 ¼ 25 C and YF;o ¼ 1, we have two equations in two unknowns: m_ 00w and m_ 00F for the extinction conditions. However, under only suppression of the burning rate, Equation (9.80) applies, giving a nearly linear result in terms of q_ 00e and m_ 00w as well as YO2 ;1 . The nonlinearity of the blocking effect can be ignored as a first approximation, and the experimental results can be matched to this theory. By subtracting Equation (9.81a) from Equation (9.80), we express the critical mass loss flux as rYF;o YF;o hc þ cp ðTv T1 Þ cp ðTf T1 Þ 1 þ YO2 ;1 00 hc YO2 ;1 hc m_ F cp =hc ¼ 00 cp em_ F cp =hc 1 r cp ðTv T1

m_ 00F;crit

(9.81b)

Ignoring the blocking factor this gives, for hc ¼ 8 W=m2 K, m_ 00F;crit ð1Þð25 kJ=gÞ þ 1:2 10 3 kJ=g K ð360 25Þ K 25=13

ð1:2 10 3 Þð1300 25Þ 1 þ 0:233 ! 2 8 W=m K ¼ ð1Þ½0:233ð13Þ ð1:2 10 3 Þð260 25Þ 1:2 J=g K 11:24 m_ 00F;crit ¼ 17:51 m_ 00F;crit ¼ 1:56 g=m2 s To have this result match the data of Magee and Reitz [19] of about 4 g/m2s at extinction requires a value for hc of about 16–20 W/m2 K – a reasonable selection over the original estimate of 8. The critical water flow rate required for extinguishment can be found as well. To match the data more appropriately, select 5.7 g=m2 s for m_ 00F corresponding to m_ 00w and q_ 00e equal to 0 from Figure 9.17. This requires the total flame net surface heat flux to be, for L ¼ 1:6 kJ=g, q_ 00f;net ¼ ð5:7 g=m2 sÞð1:6 kJ=gÞ ¼ 9:12 kW=m2

THE BURNING RATE OF COMPLEX MATERIALS

267

Therefore, if this flame heat flux does not appreciably change, reasonable for a vertical surface, from Equation (9.80) m_ 00w Lw ¼ q_ 00f;net þ q_ 00e m_ 00F L or 9:12 þ q_ 00e ð1:56 g=m2 sÞð1:6 kJ=gÞ 2:6 kJ=g ¼ 0:38 q_ 00e þ 2:55

m_ 00w crit ¼

Accordingly, at a water application of 5.2 g/m2 s (0.0076 gpm/ft2), the critical (minimum) external flux to allow the flame to survive is q_ 00e ¼

5:2 2:55 ¼ 7:0 kW=m2 0:38

This is roughly 0.23 cal/cm2 s or 9.6 kW/m2, suggested from the data of Magee and Reitz. It demonstrates that the theory does well at representing extinction. It should also be noted that the range of water flux for extinguishment is roughly 10–100 times smaller than required by common sprinkler design specifications. The Magee application rates are about 0.005 gpm/ft2 while common prescribed sprinkler densities range from 0.05 gpm/ft2 (light) to 0.5 gpm/ft2 (extremely hazardous).

9.8 The Burning Rate of Complex Materials A burning rate of common materials and products in fire can only be specifically and accurately established through measurement. Estimates from various sources are listed in Table 9.4. Also, oxygen consumption calorimeters are used to measure the energy release rate of complex fuel packages directly. In most cases, the results are a combination of effects: ignition, spread and burning rate.

Table 9.4

Estimates of burning rates for various fuel systems [20] g/s

Waste containers Waste containers Upholstered chairs (single) Upholstered sofas Beds/mattresses Closet, residential Bedroom, residential Kitchen, residential House, residential

(office size) (tall industrial)

2–6 4–12 10–50 30–80 40–120 40 130 180 4 104

268

BURNING RATE

The burning of wood and charring materials are not steady, as we have portrayed in our theory up to now. Figure 9.4 shows such burning behavior in which 1 m_ 00F pﬃ t

ðt is timeÞ

as the charring effects become important. However, the average or peak burning under flaming conditions has been described for wood simply by m_ 00wood ¼ Cw D n

ð9:82Þ

where n has been found empirically to be about 12 and D is the diameter or equivalent thickness dimension of a stick. This result is empirical with Cw as 0:88 mg=cm1:5 s for Sugar pine and 1.03 mg/cm1.5 s for Ponderosa pine [21]. Even a Cw of 1.3 was found to hold for high density (0.34 g=cm3 ) rigid polyurethane foam [22]. Although Equation (9.82) appears to hold for a single stick, it is not necessarily the case. These charring materials with an effective B of roughly ð0:233Þð12Þ=6 0:46 cannot easily sustain flaming without an external heat flux. In wood fuel systems, this is accomplished by the arrangement of logs in a fireplace, the thermal feedback in room fires or in furniture/pallet style arrays, known as ‘cribs’. A wood crib is an ordered array of equal sized wood sticks in equally spaced layers. The sticks in the wood burn either according to Equation (9.82) and the exposed stick surface area ðAÞ or by the heat flux created within the crib by the incomplete combustion of pyrolyzed fuel in a low porosity crib. The average gas temperature in the crib is related to the air mass flow rate for this ventilation-limited latter case. In that case, the crib average-peak flaming mass loss rate has been experimentally found to be proportional to pﬃﬃﬃﬃ m_ air air Ao H where H ¼ height of the crib Ao ¼ cross-sectional area of the vertical crib shafts Thus, it is found that the time-average burning rate can be represented by pﬃﬃﬃﬃ pﬃﬃﬃﬃﬃﬃ air Ao H Ao sD m_ crib ¼f ¼f

1=2

1=2 A Cw D Cw D A A

ð9:83Þ

where s is the spacing between the sticks. This is shown in Figure 9.18 by the correlation established by Heskestad [23] and compared to laboratory data taken by students in a class at the University of Maryland [24]. Note that for a tall crib its height could be a factor according to the theory, but it is not included in the correlation.

CONTROL VOLUME ALTERNATIVE TO THE THEORY OF DIFFUSIVE BURNING

Figure 9.18

269

Burning rate correlation for wood cribs according to Heskestad [23] with student data [24]

9.9 Control Volume Alternative to the Theory of Diffusive Burning This chapter began by discussing the steady burning of liquids and then extended that theory to more complex conditions. As an alternative approach to the stagnant layer model, we can consider the more complex case from the start. The physical and chemical phenomena are delineated in macroscopic terms, and represented in detailed, but relatively simple, mathematics – mathematics that can yield algebraic solutions for the more general problem. The approach is to formulate the entire burning problem using conservation laws for a control volume. The condensed phase will use control volumes that move with the vaporization front. This front is the surface of a regressing liquid or solid without char, or it is the char front as it extends into the virgin material. The original thickness, l, does not change. While the condensed phase is unsteady, the gas phase, because of its lower density, is steady or quasi-steady in that its steady solution adjusts to the instantaneous input of the condensed phase. The condensed phase (solid or liquid) is considered as a general phase that will vaporize to a gaseous fuel with a mass fraction YF;o , and can possibly form char with properties designated by ðÞc. The virgin fuel properties are designated without any subscript. Water is considered and is sprayed into the system, some reaching the surface (1-Xw) and some being evaporated in the flame (Xw ¼ fraction of water evaporated in the flame). The water spray considered is that which is fully evaporated either in the flame or on the surface. Of course in reality some water applied never reaches the fuel, and this must ultimately be considered as an efficiency factor. Therefore the water considered is that which completely contributes to suppression. The water is applied per unit area of the condensed phase surface and is assumed not to affect the flow rate of gaseous fuel from the surface. Any blockage effect of the water, or any other extinguishing agent that might be considered, is an effect ignored here, but might be included in general. This effect

270

BURNING RATE

must be considered for agents that act by blocking the surface as well as absorbing energy. The control volume relationships that are considered are based on a moving control volume surface with velocity, w, while the medium velocity is v. The conservation of mass is given as ZZZ ZZ d dV þ ðv wÞ n dS ¼ 0 ð9:84Þ dt CV

CS

The conservation of species is ZZZ ZZ ZZZ d Yi dV þ Yi ðv wÞ n dS ¼ ð m_ 000 F Þsi dV dt CV

CS

ð9:85Þ

CV

where Yi is the mass fraction of species i, m_ 000 F is the consumption rate of fuel per unit volume and si is the stoichiometric coefficient for the ith species per unit mass of fuel reacted. The conservation of energy is based on the sensible enthalpy per unit mass of the medium, hð¼ cp T, based on Equation (3.40)): d dt

ZZ Z CV

Z Z Z ZZ ZZ dp 000 hdV þ hðv wÞ ndS V ¼ m_ F hc dV q_ 00 ndS dt CS

CV

ð9:86Þ

CS

where the pressure term is neglected in the following since the pressure is constant, except for minor hydrostatic variations, in both the condensed and the gas phases. The unit normal, n, is taken as positive in the outward direction from the control volume surface, S. Hence the last term in Equation (9.86) is the net heat added to the control volume. The specific geometry of the problem is illustrated in Figure 9.19. The freestream is designatated by ðÞ1 and the backface by ðÞo. The char front, or surface regression if no char, velocity is given by dc =dt. First we will consider the condensed phase and three control volumes: the char, the vaporizing interfacial surface and the virgin material. Two variations of the virgin material control volume will be used: one where the process is steady within the material and there is no influence of the backface (thermally thick) and the other where the backface has an influence (thermally thin). A separate control volume will be considered for the water in which liquid water arrives at the control volume and is converted into vapor at the boiling point of water, Tb . The water is assumed to evaporate

Figure 9.19

General gas phase and solid phase configuration

CONTROL VOLUME ALTERNATIVE TO THE THEORY OF DIFFUSIVE BURNING

Figure 9.20

271

Control volumes selected in the condensed phase

in a steady process. Finally, a control volume will be considered for the gas phase whose thickness is representative of a constant boundary layer or flame zone.

9.9.1

Condensed phase

The control volumes for the condensed phase are shown in Figure 9.20. The control volumes are selected to include the water evaporated on the surface that is depicted as a region, but is assumed uniformly distributed over the surface. The surface area is really a differential, small area. The char control volume only contains the char at a constant density, c . Similarly, the virgin material has a constant density, . The interfacial control volume has zero volume, and it is assumed that the transition from the condensed phase to the gaseous fuel (and char) occurs quickly in this small region. It is generally found that for high heat flux conditions in fire, this is a good assumption, but it is recognized that this process is distributed over a region. In the virgin solid, there are two alternative control volumes: (a) the entire region, which includes the effect of the backface, and can represent a thermally thin solid, or (b) the region bounded by c and s , which both move at the same speed dc =dt. The material within this volume is fixed and unchanged with time as long as s does not reach the backface. As long as this condition is met, the solid can be considered thermally thick. The time t over which this condition is met is fulfilled pﬃﬃﬃﬃﬃ when l c t, where is the thermal diffusivity. It should be noted that the surfaces coinciding with c are moving control surfaces with w ¼ dc =dt j, where j is

272

BURNING RATE

positive downward in the y direction. The details follow. The symbol notation is that used throughout the text. The analysis follows: Conservation of energy for the char Z X X d c c hc dy þ hðv wÞ n ¼ q_ 00added dt 0 CS CS Z c @ dc dc ðc hc Þdy þ c hc ðTv Þ þ c hc ðTv Þ 0

dt dt 0 @t þ m_ 00F ðhg ðTs Þ hg ðTv ÞÞ ¼ q_ 00net q_ 00k q_ 00w

ð9:87Þ

Conservation of energy for the water 0 þ ð1 Xw Þm_ 00w hw;g ðTb Þ ð1 Xw Þm_ 00w hw ðT1 Þ ¼ q_ 00w

ð9:88Þ

The enthalpies can be expressed in terms of the heat of vaporization of the water: hw;g ðTb Þ hw ðT1 Þ ¼ cw;lig ðTb T1 Þ þ hw;fg Lw

ð9:89Þ

where hw;fg is the heat of vaporization. Conservation of energy for the vaporization interface 0 þ m_ 00F hg ðTv Þ þ c

dc dc 0

ð 1Þhc ðTv Þ þ 0

ðþ1ÞhðTv Þ ¼ q_ 00k q_ 00k;o dt dt

ð9:90Þ

Conservation of energy for the virgin solid including the backface d dt

Z

l c

dc h dy þ 0

ð 1ÞhðTv Þ þ 0 ¼ q_ 00k;o q_ 00o dt

ð9:91Þ

Differentiating the integral realizing c depends on time, Z

l

c

@ ðhÞ dy ¼ q_ 00k;o q_ 00o @t

Conservation of mass for the vaporizing interface 0þ

m_ 00F

dc dc þ c 0

ð 1Þ þ 0

ðþ1Þ ¼ 0 dt dt

or m_ 00F ¼ ð c Þ

dc dt

ð9:92Þ

CONTROL VOLUME ALTERNATIVE TO THE THEORY OF DIFFUSIVE BURNING

273

Conservation of energy in the virgin solid for steady, thermally thick conditions l c

pﬃﬃﬃﬃﬃ t

In this case there is no heat transfer at y ¼ s : d dt

Z

s

c

dc dc h dy þ 0

ð 1ÞhðTv Þ þ 0

ðþ1ÞhðTo Þ ¼ q_ 00k;o dt dt

ð9:93Þ

Combining with Equation (9.92) and defining the char fraction, Xc ¼ c = gives m_ 00F cðTv To Þ ¼ q_ 00k;o 1 Xc where c is the specific heat of the solid. The above equations are combined to form the condition for the burning rate at the surface, y ¼ 0. This becomes a boundary condition for the gas phase where combustion is occurring. This is done best by adding Equations (9.87), (9.88), (9.90) and (9.91). In doing so, a grouping of the enthalpies at the vaporization interface arises which represents the enthalpy of vaporization defined in terms of energy per unit mass of the virgin solid (by convention): hpy ð1 Xc Þhg ðTv Þ þ Xc hc ðTv Þ hðTv Þ

ð9:94Þ

The final result for the burning rate can be expressed in terms of the heat of gasification of the virgin solid, Lo : Lo hpy þ cðTv To Þ

ð9:95Þ

The net surface heat flux is given explicitly as 4 q_ 00net ¼ q_ 00f;c þ q_ 00f;r þ q_ 00e;r ðTs4 T1 Þ

ð9:96Þ

where q_ 00f;c ¼ flame convective heat flux q_ 00f;r ¼ flame radiative heat flux q_ 00e;r ¼ the external raditive heat flux 4 ðTs4 T1 Þ ¼ re-radiative loss Z c @ m_ 00F Lo 00 00 00 4 4 ðc hc Þdy ð1 Xw Þm_ 00w Lw ¼ q_ f;c þ q_ f;r þ q_ e;r ðTs T1 Þ

@t 1 Xc 0 00 Z l @ m_ F cðTv To Þ 00 ðhÞdy q_ o þ

1 Xc c @t

ð9:97Þ

274

BURNING RATE

where the quantity in the [ ] is zero for the steady, thermally thick solid case. If the virgin solid is thick enough, it can equilibrate to the steady state, but the increase in the surface temperature as the char layer thickens is a continuing unsteady effect. The unsteady term involving the char enthalpy will not be very large since the char density is small. The decrease in the burning rate for the charring case is apparent, especially considering the surface temperature increase over Tv to Ts and the reduction due to the char fraction. One can also see why the experimentally measured time-average, effective heat of gasification of charring materials (L) from mass loss or calorimetry apparatuses is higher than that of noncharring solids, i.e. L ¼ Lo =ð1 Xc Þ, Equation (9.7).

9.9.2

Gas phase

The control volume in the gas phase (Figure 9.19) is considered as stationary, steady and responds fast to any changes transmitted from the condensed phase. A portion of the water used in suppression is evaporated in the flame. The flame is a thickness of R within the boundary layer. Kinetic effects are important in this region where essentially all of the combustion occurs. The conservation relationships follow: Conservation of mass m_ 00P ¼ m_ 00F þ m_ 001 þ Xw m_ 00w þ ð1 Xw Þm_ 00w

ð9:98Þ

emphasizing the separate contributions of water vapor due evaporation from the flame and the surface. Conservation of species The chemical equation is represented in terms of the stoichiometric mass ratio of oxygen to fuel reacted, r: 1 g F þ r g O2 ! ðr þ 1Þ g P

ð9:99Þ

(If this reaction were in a turbulent flow, excess oxidizer will be present due to the ‘unmixedness’ of the fuel and oxygen in the turbulent stream. Hence, r will effectively be larger for a turbulent reaction than the molecular chemical constraint.) Assuming a uniform reaction rate in the flame, we obtain:

YF;o m_ 00F ¼ m_ 000 F R 00 Oxygen :

Yox;1 m_ 1 ¼ r m_ 000 F R

Fuel :

ð9:100Þ ð9:101Þ

Kinetics Arrhenius kinetics apply and these are represented as

E=ðRTÞ m_ 000 F ¼ Ae

ð9:102Þ

The thickness of the flame zone can be estimated in a manner similar to that used for the premixed flame. A control volume is selected between the condensed phase surface and the point just before the reaction zone. This is the ‘preheat zone’, which is heated to Ti .

275

CONTROL VOLUME ALTERNATIVE TO THE THEORY OF DIFFUSIVE BURNING

Heat transfer is received in this zone from the flame by conduction. The result is m_ 00F cp ðTi Ts Þ ¼ kðTf Ti ÞR

ð9:103Þ

Equation (9.103) can give an estimate for the flame thickness that is similar to that of the premixed flame by using Equation (9.100) and an estimate of 3 for the temperature ratio: R

3kYF;o cp m_ 000 F

1=2 ð9:104Þ

We will avoid the kinetics subsequently in solving for the burning rate and the flame temperature, but it is important in understanding extinction. The solutions for the burning rate and flame temperature are quasi-steady solutions for the gas phase. Where steady conditions do not exist, we have extinction ensuing. Extinction will be addressed later in Section 9.10. Conservation of energy The products leave the flame at the flame temperature, Tf , and water vapor flows in from the water evaporated on the surface and water evaporated in the flame. The control volume excludes the water droplets in the flame which receive heat, q_ 00f;w , from the flame control volume. As computed in Equation (9.88) for the fraction of water evaporated now in the flame, q_ 00f;w ¼ Xw m_ 00w hw;g ðTb Þ þ Xw m_ 00w hw ðT1 Þ ¼ Xw m_ 00w Lw

ð9:105Þ

The energy balance for the flame control volume is given as m_ 00P cp Tf m_ 001 cp T1 Xw m_ 00w cp Tb m_ 00F cp Tf ð1 Xw Þm_ 00w cp Tb _ 000 ¼ q_ 00f;c q_ 00f;w Xr m_ 000 F R hc þ m F R hc or m_ 00P cp Tf m_ 001 cp T1 m_ 00w cp Tb m_ 00F cp Tf ¼ q_ 00f;c Xw m_ 00w Lw þ ð1 Xr Þm_ 000 F R hc ð9:106Þ In this case, all of the radiation loss from the flame is accounted for by Xr, including that received by the surface of the condensed phase. Therefore flame radiation heat flux does not show up here. Equation (9.106) can be modified by eliminating all of the mass terms in favor of the burning rate (flux) term. This gives rYF;o m_ 00F cp ðTf T1 Þ cp ðTs T1 Þ þ cp ðTf T1 Þ Yox;1 00 m_ þ w00 ½cp ðTf T1 Þ cp ðTb T1 Þ þ Xw Lw ð1 Xr ÞYF;o hc m_ F ¼ q_ 00f;c

ð9:107Þ

276

BURNING RATE

This equation along with Equation (9.97) allows us to eliminate the convective flame heat flux to develop an equation for the flame temperature. This equation will still contain the burning rate in terms of the effective heat of gasification, Lm . From Equation (9.97), we define Lm as the ‘modified’ heat of gasification by the following: q_ 00f;c ¼ m_ 00F Lm Lo 1 4 m_ 00F

00 q_ 00f;r þ q_ 00e;r ðTs4 T1 Þ: 1 Xc m_ F Z c @ ðc hc Þ dy ð1 Xw Þm_ 00w Lw

0 @t 00 Z l @ m_ F cðTv To Þ 00 ðhÞ dy q_ o þ

1 Xc c @t

(9.108)

This shows that this modified heat of gasification includes all effects that augment or reduce the mass loss rate. Recall that the term in the [ ] becomes zero if the solid is thermally thick and the virgin solid equilibrates to the steady state. Equating Equations (9.107) and (9.108) gives an equation for the flame temperature:

1 cp ðTf T1 Þ ¼ 1 þ ðrYF;o =Yox;1 Þ þ ðm_ 00w =m_ 00F Þ 00 Lo m_ w þ cp ðTs T1 Þ

ð1 Xr ÞYF;o hc

½Lw þ cp ðTb T1 Þ 1 Xc m_ 00F (9.109) Z c 00 00 4 4 ½q_ f;r þ q_ e;r ðTs T1 Þ

ð@=@tÞðc hc Þ dy 0

þ

½m_ 00F cðTv To Þ=ð1 Xc Þ

Rl

_ 00o c ð@=@tÞðhÞ dy q

m_ 00F

This exactly reduces to the stagnant layer one-dimensional pure diffusion problem (Equation (9.56)). In the stagnant layer solution, an accounting for the water evaporated in the flame and the radiation loss would have modified the energy equation as (Equation (9.23)). cp m_ 00F

dT d2 T _ 000

k 2 ¼ m_ 000 F ð1 Xr Þhc Xw m w Lw dy dy 00 m_ w 000 ¼ m_ F ð1 Xr Þhc

Xw Lw m_ 00F

ð9:110Þ

The term in the bracket can be regarded as an equivalent heat of combustion for the more complete problem. If this effect is followed through in the stagnant layer solution of the ordinary differential equations with the more complete boundary condition given by

GENERAL CONSIDERATIONS FOR EXTINCTION BASED ON KINETICS

277

Equation (9.108), then the B number would become B¼

Yox;1 ½ð1 Xr Þhc ðm_ 00w =m_ 00F ÞXw Lw =r cp ðTs T1 Þ Lm

(9.111)

The burning rate is then given as m_ 00F ¼

hc lnð1 þ BÞ cp

ð9:112aÞ

or m_ 00F cp hc ¼ e 1 B

ð9:112bÞ

m_ 00F cp x hc x ¼ B k k e 1

ð9:112cÞ

and

Alternatively, in Equation (9.108), the convective heat flux could be approximated as q_ 00c ¼ hc ðTf Ts Þ

ð9:113Þ

and we obtain an approximate result for the burning rate using Equation (9.109). In summary, Equations (9.111) and (9.112) represent a more complete solution for the burning rate, and Equation (9.109) gives the corresponding flame temperature.

9.10 General Considerations for Extinction Based on Kinetics Let us reconsider the critical flame temperature criterion for extinction. Williams [25], in a review of flame extinction, reports the theoretical adiabatic flame temperatures for different fuels in counter-flow diffusion flame experiments. These temperatures decreased with the strain rate (u1 =x), and ranged from 1700 to 2300 K. However, experimental measured temperatures in the literature tended to be much lower (e.g. Williams [25] reports 1650 K for methane, 1880 K for iso-octane and 1500 K for methylmethracrylate and heptane). He concludes that 1500 50 K can represent an approximate extinction temperature for ‘many carbon–hydrogen–oxygen fuels burning in oxygen–nitrogen mixtures without chemical inhibitors’. Macek [27] examined the flammability limits for premixed fuel–air systems and small diffusion flames under natural convection conditions, and computed the equilibrium flame temperature for these flame systems. Data were considered for the alkanes and alcohols at their measured premixed lower flammability limits, and at their measured

278

BURNING RATE

Figure 9.21

Adiabatic flame temperatures at extinction for premixed and diffusion flames (from Macek [26])

oxygen concentration for extinction in the standard limiting-oxygen index (LOI) method. The LOI method is a small-scale test method for diffusion flames in an inverted hemispherical burner arrangement within a chamber of controlled oxygen. Figure 9.21 shows the computed adiabatic flame temperatures at these limit conditions for complete combustion and for the alternative extreme condition of burning only to CO and water. Macek argues that extinction of the diffusion flames should occur somewhere between the two limits of stoichiometry, and therefore the implication is that the critical flame temperature is about 1600 K, as generally suggested for the premixed flames in the literature. This criterion is often applied for premixed flames in which the adiabatic flame temperature for constant specific heat follows from Equation (4.55) as cp ðTf;ad T1 Þ ¼ YF;o hc

ð9:114Þ

where the fuel mass fraction supplied is YF;o . The corresponding expression for a diffusion flame is more complicated, as shown by Equation (9.109). The LOI conditions correspond to, at least, near-stoichiometric fuel concentrations as characteristic of a diffusion flame, and hence the temperatures for extinction are higher than those for the LFL premixed flames in Figure 9.21. The value of 1600 K appears to be consistent with the results in Figure 9.21, and suggests that premixed flames may burn more completely than diffusion flames at extinction. Since the heat transfer processes are not influenced by whether the system is premixed or not, the same kinetics should apply to both flame configurations. Flames in natural fire conditions are affected by gravity-induced flows that are relatively low speed, with their inherent turbulent structure. Therefore, it might be assumed that the flow field has a small effect on natural convection flames, and the critical extinction temperature is more consistent with the 1200–1300 C suggested by Williams [25]; we will use 1300 C ( 1600 K) in the following for diffusion flames. Indeed, Roberts and Quince [27] concluded 1600 K is an approximate exinction limit as they successfully predicted the liquid temperature needed for sustained burning (fire-

279

GENERAL CONSIDERATIONS FOR EXTINCTION BASED ON KINETICS

point). They equated bOF and bOT and used the Clausius-Clapeyron equation along with this critical flame temperature.

9.10.1

A demonstration of the similarity of extinction in premixed and diffusion flames

The control volume analysis of the premixed flame of Section 4.5.4 can be used together with the analysis here in Section 9.9 for the diffusion flame to relate the two processes. We assume the kinetics is the same for each and given as in Equation (9.102). Since we are interested in extinction, it is reasonable to assume the heat loss from the flame to be by radiation from an optically thin flame of absorption coefficient, : 4 ÞR q_ 00loss ¼ 4ðTf4 T1

Then for the premixed case, following Equation (4.42), 000 m_ F 4 Þ ½YF;o hc cp ðTf To Þ ¼ 4ðTf4 T1 YF;o

ð9:115Þ

ð9:116Þ

The analogous equation can be derived for the diffusion flame. Here, we will consider, for simplicity, no application of water and no charring. The optically thin flame heat loss allows an explicit expression for 4 Xr m_ 00F hc ¼ 4ðTf4 T1 ÞR

ð9:117Þ

and half of this, due to symmetry, is the incident surface flame radiation, q_ 00f;r . Hence, from Equations (9.100) and (9.107), it follows that 000 rYF;o m_ F R 4 ÞR YF;o hc 1 þ cp ðTf T1 Þ þ cp ðTs T1 Þ ¼ q_ 00f;c þ 4ðTf4 T1 YF;o Yox;1 ð9:118Þ By substituting for the flame convective heat flux from Equation (9.108) for steady conditions and no external heat flux (and no char and water), the analogue to Equation (9.116) follows for the diffusion flame:

¼

2ðTf4

m_ 000 F YF;o

rYF;o YF;o hc L 1 þ cp ðTf T1 Þ þ cp ðTs T1 Þ Yox;1

4 T1 Þ

4 ðTv4 T1 Þ þ R

ð9:119Þ

The right-hand side (RHS) of Equations (9.116) and (9.119) represent the net heat loss and the left-hand side (LHS) represents the energy gain. The gain and the loss terms can be plotted as a function of the flame temperature for both the diffusion and premixed flames as ‘Semenov’ combustion diagrams. Intersection of the gain and loss curves indicates a steady solution, while a tangency indicates extinction.

280

BURNING RATE

Figure 9.22

(a) Computed steady state solutions for premixed n-heptane propagation. (b) Computed critical condition (LFL) for heptane propagation

To illustrate these Semenov diagrams in reasonable quantitative terms, n-heptane was selected with the following properties: hc ¼ 41:2 kJ=g ðTable2:3Þ r ¼ 3:15 Tv ¼ 98 C ðTable 9:2Þ k ¼ 0:117 W=m K L ¼ 0:477 kJ=g cp ¼ 1:3 J=g K o ¼ 1:18 kg=m3 To ¼ T1 ¼ 25 C E=R ¼ 19 133 K

ðfrom reference ½25Þ

A ¼ 1:41 108 g=m3 s The A value was selected to match the lower flammability limit of YF ¼ 0:0355 using Equation (4.40). The computed results are shown in Figures 9.22 and 9.23. The figures show the steady state solutions for the resulting flame temperatures. For premixed propagation, the computed temperature at stoichiometric fuel conditions is nearly the adiabatic flame temperature due to the steep nature of the gain curve. As the fuel concentration drops, so does the flame temperature. In Figure 9.22(b), it is clearly shown that at a fuel concentration of 0.035, there is a tangency of the loss and gain curves, which indicates a critical condition below which there is no longer a steady solution. This is the computed LFL and agrees well with the experimental value. The figure also clearly shows that, in general, two steady solutions are possible (S and U), but that the upper one is stable (S) and the lower one is unstable (U). The critical temperature for no sustained

APPLICATIONS TO EXTINCTION FOR DIFFUSIVE BURNING

Figure 9.23

281

(a) Computed steady burning for n-heptane as a function of the ambient oxygen mass fraction. (b) Critical conditions at extinction

propagation is about 1050 C, while the corresponding adiabatic flame on the abscissa is about 1150 C. This is in contrast to the value of 1300 C commonly assumed and indicated earlier. However, it should be emphasized that the computation is approximate, and is especially subject to the constant value for specific heat used in the calculation. Similar results are computed for the diffusion flame under steady burning of n-heptane liquid. The resulting flame temperatures are given in Figures 9.23(a) and (b). The corresponding burning rates could be determined from Equation (9.112). Again, at ambient air conditions (0.233), the flame temperature is similar to the adiabatic flame temperature. As the ambient oxygen concentration drops, the flame temperature is reduced until a critical value, c, or about 950 C results. Its corresponding adiabatic flame temperature is about 1100 C. Again these results are approximate. However, they do show that the critical temperatures for extinction of premixed and diffusion flames are similar here for the heptane. Hence, the empirical critical value of about 1300 C is credible.

9.11 Applications to Extinction for Diffusive Burning Let us examine the 1300 C criterion as a condition of flame extinction in diffusive burning to see how these critical conditions depend on oxygen and external radiation. After all, these two parameters – oxygen and radiation – are the two significant variables that make most solids burn. We will look to some data for anchoring this application. Tewarson and Pion [18] conducted tests on the burning of materials in the FMRC flammability apparatus. Horizontal samples of 60–100 cm2 were tested in order to measure their steady burning rate under radiant heating and varying oxygen concentrations as fed by O2/N2 steady flows of 50–70 L/min or speeds of about 0.5 cm/s. These speeds constitute basically natural convection burning conditions, and a heat transfer coefficient of about 9 W/m2 K was estimated in applying the theoretical results of Section 9.10. The coefficient is not precisely known for the apparatus, while contributes to some uncertainty in the results. They determined the oxygen concentration to cause extinction,

282

BURNING RATE Table 9.4

Material Heptane Styrene PMMA Methyl alcohol Polyoxymethylene (POX) Wood (Douglas fir)

Material properties

hc (kJ/g)

L (kJ/g)

Tv ( C)

r (g O2/g fuel)

41.2 27.8 24.2 19.1 14.4 12.4

0.477 0.64 1.6 1.195 2.43 ~6.8

98 146 397 64.5 425 382

3.17 2.14 1.86 1.51 1.107 0.95

and did this for several external radiant heating conditions in some cases. The latter was done for PMMA in some detail. Computations based on Section 9.10 were performed for some of the materials considered by Tewarson and Pion and will be described here [28]. The following assumptions are made in applying the extinction theory: 1. The burning rate is relatively small, so the B number is small and lnð1 þ BÞ ¼ B. 2. The flame radiation is taken as zero since the flame size diminishes. 3. The fuel is pure so YF;o ¼ 1. 4. The flame temperature at extinction is 1300 C. 5. The heat transfer coefficient is taken at 9 W/m2 K. 6. The ambient temperature is 25 C unless otherwise specified. 7. The properties needed are taken from Reference [28] for consistency and are given in Table 9.4. The heat of gasification of wood is approximated since it is unsteady and charring. In order to apply this extinction criterion, it will further be assumed that the steady state form of Equation (9.112) applies with no char, no water and no backface loss. As a consequence the burning rate can be described as 1 hc Yox;1 hc 00 00 4 4 m_ F ¼ ð9:120Þ þ q_ e;r ðTv T1 Þ L cp r cp ðTv T1 Þ Apply Equation (9.109) for these pure materials under conditions as steady, no char and no water: cp ðTf T1 Þ ¼

4 hc L þ cp ðTv T1 Þ þ ½q_ 00e;r ðTv4 T1 Þ=m_ 00F 1 þ ðr=Yox;1 Þ

ð9:121Þ

Eliminate the burning rate from Equations (9.120) and (9.121) in order to obtain the following relationship between external radiation, oxygen and the flame temperature: ðhc =cp Þ ðYox;1 hc =rÞ cp ðTv T1 Þ ðYox;1 =rÞ 1 ¼ L hf½ðYox;1 hc =r cp ðTv T1 Þ ½1 þ r=Yox;1 cp ðTf T1 Þg 4 Þgi fðhc =cp Þ ðYox;1 hc =rÞ cp ðTv T1 Þ þ q_ 00e;r ðTv4 T1

ð9:122Þ

Oxygen mole fraction at extinction

APPLICATIONS TO EXTINCTION FOR DIFFUSIVE BURNING

283

0.25 0.2 0.15 0.1 Experiment

0.05 0 0

100

Theory 200

300

400

Ambient Temperature ºC

Figure 9.24

Oxygen index for polystyrene as a function of ambient temperature [18, 27]

In the denominator, the first term in the braces is the net energy above the minimum value to sustain the flame and the second term in the braces is the net heat flux to the fuel. Both of these terms must be positive for a valid physical solution. For no external radiant heating, and for all of the following terms as small: 4 ðTv4 T1 Þ, cp ðTv T1 Þ, Yox;1 =r, then Equation (9.122) can be solved to give an approximate solution for the critical oxygen concentration: Yox;1

cp ðTf T1 Þ 1:53 ¼ ðhc LÞ=r 13:1 L=r

ð9:123Þ

This gives the approximate oxygen mass fraction at extinction. For liquid fuels, where typically L < 1 kJ=g, the lowest the critical oxygen mass fraction can be is 0.12, or 0.11 as a mole fraction. Also, this approximate result explains why the critical oxygen concentration increases with L and decreases with the ambient temperature. Figure 9.24 gives experimental results for polystyrene, compared to the theoretical value based on the criterion using Equation (9.122) [28]. Equation (9.122) gives a general relationship between the critical extinction relationship and the external heat flux and oxygen concentration at 25 C. Experimental results are compared to the theory of Equation (9.122) for polymethyl methracrylate (PMMA) in Figure 9.25. The heat transfer coefficient has a small influence on the theoretical results, but a more interesting effect is a singularity at a mass fraction of about 0.12. This occurs when the energy needed to sustain the flame is just equal to the energy from combustion, i.e. the first brace in the denominator of Equation (9.122) is zero. This singularity is approximately given as Yox;1

cp ðTf T1 Þ ¼ 0:12 hc =r

ð9:124Þ

This singularity is independent of the fuel. Figure 9.26 shows results over a complete range of oxygen concentrations for the fuels listed in Table 9.4. It shows that Douglas fir requires about 18 kW/m2 to burn in air, while most of the other fuels burn easily in air with no external heating. Heptane will continue to burn with no external heating to the asymptotic oxygen fraction of 0.12. Below an oxygen fraction of 0.18, external heating is needed to sustain PMMA.

284

BURNING RATE

Figure 9.25

Critical heat flux and oxygen concentration at extinction for PMMA [18, 28]

The critical burning mass flux at extinction is between 2 and 4 g/m2 s when burning in air, as seen in the theoretical plot of Figure 9.27. A curious minimum mass flux at extinction appears to occur for the oxygen associated with air for nearly all of the fuels shown. The critical mass flux at exinction is difficult to measure accurately, but experimental literature values confirm this theoretical order-of-magnitude for air. The asymptote at 0.12 has not been verified. These theoretical renditions should be taken as qualitative for now.

Figure 9.26

Computed critical conditions for fuels burning at 25 C [28]

REFERENCES

Figure 9.27

285

Critical mass flux at extinction [28]

References 1. Hopkins, D. and Quintiere, J.G., Material fire properties and predictions for thermoplastics, Fire Safety J., 1996, 26 241–268. 2. Spearpoint, M.J. and Quintiere, J.G., Predicting the ignition and burning rate of wood in the cone calorimeter using an integral model, Masters Thesis, Department of Fire Protection Engineering, University of Maryland, 1999. 3. Tewarson, A., Generation of heat and chemical compounds in fires, in The SFPE Handbook of Fire and Protection Engineering, 2nd edn (eds P. J. DiNenno), Section 3, National Fire Protection Association Quincy, Massachusetts, 1995, p. 3–68. 4. Dillon, S.E., Quintiere, J.G. and Kim, W.H., Discussion of a model and correlation for the ISO 9705 room-corner test, in Fire Safety Science – Proceedings of the 6th International Symposium, IAFSS, 2000, p. 1015–1026. 5. Spalding, D.B., The combustion of liquid fuels, Proc. Comb. Inst., 1953, 4, pp. 847–864. 6. Incoperra, F.P. and deWitt, D.P., Fundamentals of Heat and Mass Transfer, 4th edn, John Wiley & Sons, New York, 1996. 7. Emmons, H.W., The film combustion of liquid fuel, Z. Angew Math. Mech., 1956, 36(1), 60–71. 8. Glassman, I., Combustion, Academic Press, New York, 1977, p. 185. 9. Kosdon, F.J., Williams, F.A. and Buman, C., Combustion of vertical cellulosic cylinders in air, Proc. Comb. Inst., 1968, 12, pp. 253–64. 10. Kim, J.S., deRis, J. and Kroesser, F.W., Laminar free-convective burning fo fuel surfaces, Proc. Comb. Inst., 1969, 13, pp. 949–61. 11. Ahmad, T. and Faeth, G.M., Turbulent wall fires, Proc. Comb. Inst., 1979, 17, pp. 1149–60. 12. Tien, C.L., Lee, K.Y. and Stretton, A.J., Radiation heat transfer, in The SFPE Handbook of Fire Protection Engineering, 2nd edn (eds P.J. DiNenno et al.), Section 1, National Fire Protection Aassociation, Quincy, Massechusetts, 1995, pp. 1–65 to 1–79. 13. Modak, A. and Croce, P., Plastic pool fires, Combustion and Flame, 1977 30, 251–65.

286

BURNING RATE

14. Iqbal, N. and Quintiere, J.G., Flame heat fluxes in PMMA pool fires, J. Fire Protection Engng, 1994, 6(4), 153–62. 15. Orloff, L., Modak, A. and Alpert, R.L., Burning of large-scale vertical surfaces, Proc. Comb. Inst., 1977, 16, pp. 1345–54. 16. Corlett, R.C. and Fu, T.M., Some recent experiments with pool fires, Pyrodynamics, 1966, 4, 253–69. 17. Kung, H.C. and Stavrianidis, P., Buoyant plumes of large - combustion scale pool fires, Proc. Comb. Inst., 1982, 19, pp. 905–12. 18. Tewarson, A. and Pion, R.F., Flammability of plastics. I. Burning intensity, Combustion and Flame, 1978, 26, 85–103. 19. Magee, R.S. and Reitz, R.D., Extinguishment of radiation-augmented plastic fires by water sprays, Proc. Comb. Inst., 1975, 15, pp. 337–47. 20. Quintiere, J.G., The growth of fire in building compartments, in Fire Standards and Safety (ed. A. F. Robertson), ASTM STP614, American Society for Testing and Materials, Philadelphia, Pennsylvania, 1977, pp. 131–67. 21. Block, J., A theoretical and experimental study of non-propagating free-burning fires, Proc. Comb. Inst., 1971, 13, pp. 971–78. 22. Quintiere, J.G. and McCaffrey, B.J., The Burning of Wood and Plastic Cribs in an Enclosure, Vol. I., NBSIR 80-2054, National Bureau of Standards, US Department of Commerce, Gaithersburg, Maryland, November 1980. 23. Heskestad, G., Modeling of enclosure fires, Proc. Comb. Inst., 1973, 14, p. 1021–1030. 24. Students, ENFP 620, Fire Dynamics Laboratory, University of Maryland, College Park, Maryland, 1998. 25. Williams, F.A., A review of flame extinction, Fire Safety J., 1981, 3, 163–75. 26. Macek, A., Flammability Limits: Thermodynamics and Kinetics, NBSIR 76-1076, National Bureau of Standards, US Department of Commerce, Gaithersburg, Maryland, May 1976. 27. Roberts, A.F. and Quince, B.W. A limiting condition for the burning of flammable liquids, Combustion and Flame, 1973, 20, 245–251. 28. Quintiere, J.G. and Rangwala, A.S., A theory for flame extinction based on flame temperature, Fire and Materials, 2004, 28, 387–402.

Problems 9.1

For pure convection burning of PMMA in air at 20 C and steady state, determine the mass loss rate per unit area for a vertical slab of height l1 ¼ 2 cm and l2 ¼ 20 cm. Use the gas properties for air. PMMA properties from Table 2.3 are: hc ¼ 25 kJ=g, L ¼ 1:6 kJ=g, Tv ¼ 380 C and hox ¼ 13 kJ=g O2 .

9.2

Polyoxymethylene burns in a mixture of 12 % oxygen by mass and nitrogen is present. Assume it burns purely convectively with a heat transfer coefficient of 8 W/m2 K. Find m_ 00 . Polyoxymethylene properties:

T1 ¼ 20 C

Tv ¼ 300 C

L ¼ 2:4 kJ=g hox ¼ 14:5 kJ=g O2

PROBLEMS 9.3

287

n-Octane spills on hot pavement during a summer day. The pavement is 40 C and heats the octane to this temperature. Wind temperature is 33 C and pressure is 1 atm.

n-Octane (liquid) Data: Ambient conditions are at a temperature of 33 C and pressure of 760 mmHg. Density of air is 1.2 kg/m3 and has a specific heat of 1.04 J/g K. Octane properties: Boiling temperature is 125.6 C. Heat of vaporization is 0.305 kJ/g. Specific heat (liquid) is 2.20 J/g K. Specific heat (gas) is 1.67 J/g K. Density (liquid) is 705 kg/m3. Lower flammability limit in air is 0.95 %. Upper flammability limit in air is 3.20 %. Use data from Table 6.1. The octane spill is burning in wind that causes a convective heat transfer coefficient of 20 W/m2 K. The same data as given above still apply. (a) Compute the heat of gasification of the octane. (b) Compute the B-number for purely convective burning but ignore the radiation effects. (c) Compute the steady burning rate per unit area ignoring all effects from radiation. (d) Compute the convective heat flux from the flame. (e) Extinction occurs, at a burning rate of 2 g/m2 s, due to a heat flux loss from the spill surface as the spill thickness decreases. Calculate the heat flux loss to cause the burning rate to reach the extinction condition. Radiation effects are still unimportant. 9.4

A candle burns at a steady rate (see the drawing below). The melted wax along the wick has a diameter D ¼ 0:5 mm and pyrolysis occurs over a length of lp ¼ 1 mm. Treat the wick as a flat plate of width D and a height of lp which has a convective heat transfer coefficient h ¼ 3 W=m2 K. Ignore all radiative effects.

288

BURNING RATE

Properties of the wax: Heat of combustion of liquid wax ¼ 48.5 kJ/g Stoichiometric air to fuel mass ratio ¼ 15 Density of the solid and liquid phases ¼ 900 kg/m3 Specific heat of the solid and liquid phases ¼ 2.9 J/g K Heat of vaporization ¼ 318 J/g Heat of fusion (melting) ¼ 146 J/g Temperature at melting ¼ 52 C Temperature at vaporization ¼ 98 C Diameter of the candle, as shown is 2 cm Also: Initial and ambient temperature in air ¼ 25 C

cp;g ðfor airÞ ¼ 1 J=g K kg ¼ 0:03 W=m K (a) Determine the mass burning rate in g/s. Assume the initial temperature in the wick is the melt temperature. (b)Assuming pure conduction heat transfer from the candle flame to the melted wax pool, and a thin constant melt layer on the wax, estimate the distance between the flame and the melt layer (i.e. ls ). Also, assume an effective heat transfer area based on the candlewick diameter, i.e. 0.5 mm. 9.5

A thick polystyrene wainscotting covers the wall of a room up to 1 m from the floor. It is ignited over a 0.2 m and begins to spread. Assume that the resulting smoke layer in the room does not descend below 1 m and no mixing occurs between the smoke layer and the lower air. Polystyrene properties: Density, ¼ 40 kg=m3 Specific heat, cp ¼ 1:5 J=g K Conductivity, k ¼ 0:4 W=m K Vaporization and ignition temperature, Tv ¼ 400 C Heat of combustion, hc ¼ 39 kJ=g Stoichiometric oxygen to fuel ratio, r ¼ 3 g=g Heat of gasification, L ¼ 1:8 kJ=g

PROBLEMS

289

Other information: Initial temperature, T1 ¼ 20 C Ambient oxygen mass fraction, Yox;1 ¼ 0:233 Specific heat of air, cp ¼ 1 J=g K

(a) The fire spreads to the top of the wainscotting, but does not spread laterally. What is the burning (pyrolysis) rate of the polystyrene over this region assuming a steady state. The vertical natural convection heat transfer coefficient, h ¼ 15 W=m2 K. (b) Later the fire spreads laterally to 2 m and the surface experiences a heat flux due to radiation from the heated room of 1.5 W/cm2. Using the small B-number approximation, account for all radiation effects and calculate the mass loss rate per unit area of polystyrene. 9.6

PMMA burns in air at an average m_ 00 ¼ 15 g=m2 s. Air temperature is 20 C and the heat transfer coefficient is 15 W/m2 K. Determine the net radiative heat flux to the fuel surface. What amount comes from the flame?

9.7

Calculate m_ 00 for pools of heptane burning in air having diameters D ¼ 1 cm, 10 cm and 1 m. Take hc =r ¼ 13 kJ=g, Tv ¼ 98 C, properties for air are at 20 C and use Tables 9.2 and 10.4. Use the small B-number approximation. Assume that q_ 00r;f ¼ ð1 e D ÞTf4 , where Tf ¼ 1600 K, an effective radiation temperature ¼ 12 m 1 , a flame absorption coefficient

9.8

A vertical slab of PMMA burns in air at an average mass loss rate of 15 g/m2 s. The air temperature is 20 C. Assume that the heat transfer coefficient of the slab is 15 W/m2 K. Determine the net radiative heat flux to the surface. What amount comes from the flame? PMMA properties are:

hc ¼ 25 kJ=g L ¼ 1:6 kJ=g Tig ¼ 380 C 9.9

Properties needed: Liquid n-decane: Boiling point ¼ 447 K Heat of vaporization ¼ 0.36 kJ/g Flashpoint ¼ 317 K Lower flammability limit (by volume) ¼ 0.6 % Specific heat ¼ 2.1 kJ/kg K

290

BURNING RATE Density ¼ 730 kg/m3 Heat of combustion ¼ 44.7 kJ/g Stoichiometric oxygen to fluid mass ratio ¼ 3.4 Thermal conductivity ¼ 0.5 W/m K Air: Kinematic viscosity ¼ 0.5 W/m K Density ¼ 1.2 kg/m3 Specific heat ¼ 1.0 kJ/kg K Gravitational acceleration, g ¼ 9:81 m=s2 Stefan–Boltzman constant, ¼ 5:67 10 11 kW=m2 K4 n-Decane is exposed to an atmospheric temperature of 20 C and a pressure of 1 atm Answer the following using the above data: (a) The decane is at 20 C. Determine the fuel vapor concentration at the liquid surface. Is the decane flammable under these conditions? (b) If the decane is at the lower flammability limit for an air atmosphere, compute the necessary surface temperature. Will the liquid autoignite at this temperature? (c) Compute the heat of gasification for the decane initially at 20 C and at the boiling point. (d) For convective heat transfer from air to 80 C and with a convective heat transfer coefficient (h) of 15 W/m2 K, compute the mass rate of evaporation per unit area of the decane at the temperature where it just becomes flammable; this is before combustion occurs. Assume the bulk temperature of the liquid is still at 20 C. (e) If the decane were exposed to an external radiant heat flux of 30 kW/m2, how long would it take to ignite by piloted ignition? Assume h ¼ 15 W=m2 K, T1 ¼ 20 C; also, the liquid is very deep, remains stagnant and acts like a solid. (f) For purely convective-controlled burning with h ¼ 15 W=m2 compute the burning rate per unit area under steady conditions for a deep pool of decane in air at 20 C and an atmosphere of 50% oxygen by mass at 20 C. (g) A 1.5 m diameter deep pool of decane burns steadily in air (T1 ¼ 20 C). Its radiative fraction is 0.25 of the total chemical energy release and 5 % of this radiative loss is uniformly incident over the surface of the pool. Accounting for all the radiative effects, what is the total energy release of the pool fire? The heat transfer coefficient is 15 W/m2 K. (Use the small B approximation.) (h) Decane is ignited over region xo ¼ 0:5 m in a deep tray. A wind blows air at 20 C over the tray at v1 ¼ 10 m=s. The convective heat transfer coefficient is found to be 50 W/m2 K. (i)

Assume the decane flame spreads like that over a solid surface; i.e. ignore circulation effects in the liquid. Assume pure convective burning. Compute the convective heat flux from the

PROBLEMS

291

flame to the surface and also compute the flame spread velocity just following ignition over xo . It is known that the flame length, xf , can be found by

2:7 xf v1 0:25 xo 0:89 m_ 00 ¼ 11:6 v1 pﬃﬃﬃﬃﬃﬃﬃ pﬃﬃﬃﬃﬃﬃﬃ xo 1 gxo gxo where m_ 00 ¼ mass burning rate per unit area 1 ¼ air density g ¼ gravitational acceleration ¼ kinematic viscosity 9.10

D. B. Spalding published a seminal paper, ‘The combustion of liquid fuels’ in the 4th International Symposium on Combustion (Reference [5]). In it he addressed the burning of droplets by using a porous sphere under both natural and forced convective flows. The figure below shows the burning of kerosene on a 1.5 inch diameter sphere under natural convection conditions. Spalding empirically found from his data that the burning rate per unit area could expressed as

3 1=4 gD m_ 00F Dcp ¼ 0:45 B3=4 k 2

292

BURNING RATE The symbols are consistent with Section 9.2 of the text. We are to use the stagnant layer theory of burning with suitable approximations to analyze the burning of a droplet in natural convection. The droplet, suspended and burning in air, is assumed to remain spherical with diameter, D. (a) From the stagnant layer theory in pure convective burning, develop an alternative, but similar, formula for m_ 00F Dcp =k. It is known from heat transfer results that

NuD ¼ 2 þ

0:589ðGr PrÞ1=4 ½1 þ ð0:469 PrÞ9=16 4=9

(b) A droplet of n-decane, C10H22, beginning at a diameter of 10 mm, burns in air which is at 25 C. The properties of n-decane are given below: Boiling point Liquid density Liquid specific heat Heat of vaporization Heat of combustion Stoichiometric ox/fuel mass ratio

174 C 730 kg/m3 1.2 J/g K 0.360 kJ/g 44.2 kJ/g 3.5 g/g

From your formula, compute the initial burning rate in g/s for pure convective burning. (c) The following additional information is available. The flame height in the wake region above the burning droplet can be estimated by

zf ¼ 0:23Q_ 2=5 1:02 D where zf is measured from the center of the droplet in m, Q_ is the firepower in kW and D is the diameter in m. The flame absorption coefficient, , can be estimated as 0.45 m 1. The mean beam length of the flame relative to its base can be expressed as

Le ¼ 0:65ð1 e 2:2 zf =D Þ D The overall radiation fraction, Xr , is 0.15. Compute the initial burning rate, now including radiation effects. Show all your work and clearly indicate any assumptions. (d) The droplet burns until no liquid fuel remains. How long does this take? Use the most complete burning rate description. Show how you include the variation of D over time. An integration process is need here. 9.11

A pool of PMMA with a diameter of 1 m burns in air. Radiation effects are to be included. In addition to using the properties from Problem 9.1, let Xr ¼ 0:25; Tf ¼ 1500 K and ¼ 0:5 m 1 . If it is known that the critical mass loss rate for PMMA is 4 g/m2 s, calculate the flux of water (m_ 00w ) needed to extinguish this fire. If instead the oxygen is reduced, at what mass fraction will extinction occur?

9.12

Polyoxymethylene is burning in the upper smoke layer of a room fire. The critical mass loss rate per unit area at extinction is known to be 2 g/m2 s for polyoxymethylene. The smoke layer temperature is 700 C.

PROBLEMS

293

Properties: Heat of gasification ¼ 2.43 kJ/g Heat of combustion ¼ 15.5 kJ/g Convective heat transfer coefficient for the configuration ¼ 15 W/m2 K Specific heat of gas layer ¼ 1 kJ/kg K Heat of combustion per unit mass of oxygen consumed ¼ 14.5 kJ/g O2 Vaporization temperature ¼ 250 C Assume purely convective burning and negligible radiation. (a) Calculate the oxygen mass fraction to cause extinction. (b) For 1 m2 of surface burning, calculate the energy release rate if the layer smoke oxygen mass fraction is 0.20. 9.13

For a standard business card, in the vertical length dimension, determine the steady burning rate (g/s) for one side of the card saturated with ethanol. Only the ethanol burns. Show your analysis and all assumptions. This is a calculation, not an experimental determination, though experiments can be conducted. State all data and sources used. You will have to make approximations and estimates for quantities in your analysis.

9.14

A polypropylene square slab, 0.3 m on a side, burns in a steady manner. Polypropylene does not char, and it can be considered that its radiation characteristics do not vary, with the flame radiative fraction at 0.38 and the flame incident heat flux to the fuel surface at 25 kW/m2. It burns in a wind as shown in various conditions as specified. Its convective heat transfer coefficient can be taken as 50 W/m2 K. The properties of polypropylene are listed as follows.

Properties: Heat of combustion, hc ¼ 38 kJ=g Heat of gasification, L ¼ 3:1 kJ=g Vaporization temperature, Tv ¼ 330 C Any other data or properties should be cited from the text. Assume that the B-number or m_ 00F cp =hc is small for your calculations in all of the following. (a) If it burns in air at an ambient temperature of 10 C, compute the burning flux in g/m2 s.

294

BURNING RATE (b) It is placed in a flow-through oven with the same wind conditions. The air temperature is 60 C. The oven radiates to the burning slab with 45 kW/m2. Compute the firepower in kW. (c) How much water spray per unit area is needed to reduce the firepower by 50 % in problem (b)? (d) Now pure oxygen is used in the oven of problem (b) with the same ambient and flow and heating conditions. Compute the burning rate in g/s. (e) What is the convective heat flux in problem (d)?

9.15

PMMA is burned, in normal air at 25 C, in a horizontal configuration (30 30 cm) under an external irradiation of 20 kW/m2. You are to quantitatively evaluate the ability of dry ice (CO2) and ice (H2O) both in solid flakes at their respective freezing point (at 1 atm) to extinguish the fire. Assume steady burning with the sample originally at 25 C with a perfectly insulated bottom. At extinction you can ignore the flame radiation. Assume that all of the flakes hit the surface and ignore the gas phase effects of the extinguishment agents. Use thermodynamic properties of the CO2 and H2O, and the property data of PMMA from Table 9.2. Compute: (a) The ratio of flow rates for dry ice and water ice to the burning rate needed for extinction. (b) The critical burning rate per unit area at extinction for both ice and dry ice. (c) The flow rate of dry and ice needed to extinguish the fire.

9.16

Use the control volume of the diffusion flame shown to derive the flame temperature of the outgoing products. ⋅ m a′′

m⋅ p′′

Flame at Tf

External Radiation No mass or heat flow

Tv

m⋅ F′′

q⋅ ′′

where m_ 00a ¼ mass flux of air diffuses toward the flame m_ 00p ¼ mass flux of products diffusing away from the flame q_ 00 ¼ m_ 00 L, heat loss to the fuel m_ 00F ¼ evaporated fuel diffuses toward the flame

cg ðTf Tv Þ ¼

Yox;1 ½ðhc LÞ=r þ cg ðT1 Tv Þ 1 þ Yox;1 =r

PROBLEMS 9.17

295

Consider the following plastic materials and their properties. These materials do not char; they melt and vaporize.

Material Polymethylmethracrylate (PMMA) Polyoxy-methylene (POM) Polystyrene (PS) Polyethylene (PE)

Rate without external radiation 00 m_ F;o (g/m2 s)

L (kJ/g)

Lb (kJ/g)

Lc (kJ/g)

hbc (kJ/g)

1.61

1.63

2.0

24.2

391

300

0.31 5.8 v, 7.5 ha

2.93

2.43

—

14.4

419

—

0.22 6.2 v, 6.0 h

2.09 2.31

1.70 3.0

3.0 3.6

27 38.4

419 444

350 360

0.59 10.8 h 0.43 3.4 h

a

Tvb ( C) Tvc ( C)

Xrb

a

R. S. Magee and R. D. Reitz [19]. A. Tewarson [3]. J. G. Quintiere [1]. d Orientation: v, vertical sample; h, horizontal sample. b c

You are to use these data in the following calculations. Use the Magee data for L and the Tewarson data for Tv, since these values have some uncertainty, as indicated by their variation from the different sources. These differences can be due to variations in the measurements or in the materials. Plastics can have additives in varying amounts, yet still be listed as one polymer. For each material and burning orientation compute the following. Assume steady burning in all your calculations. (a) Based on the burning rates measured for the horizontal and vertical samples, without any additional radiant heating, find the flame convective and radiative heat flux components. The vertical sample was 7 inches wide and 14 inches high; the horizontal was 7 inches square. (b) Assuming the flame heating is constant, write expressions for the burning rates in terms of the external radiative heat flux, q_ 00e;r , and the water application flux, m_ 00w . In these experiments, all of the water reached the surface and none was absorbed in the flame. (c) Determine, and plot if you wish, the relationship between the external radiative heat flux and the water application flux at the point of extinction. At extinction, you can consider the mass flux of the burning fuel to be small. (d) Compute the burning flux just at extinction, and show how it depends on the radiative heat flux and water flux. 9.18 The following data are given for the WTC towers: Building: 110 stories, 11 ft per story 208 ft 208 ft floor space 7 lb=ft2 of floor area of wood-based furnishings

296

BURNING RATE Jet fuel, JP4: Liquid density ¼ 760 kg/m3 Heat of combustion ¼ 43.5 kJ/g Heat of gasification ¼ 0.80 kJ/g Effective vaporization temperature ¼ 230 C Radiative loss fraction ¼ 0.20 Wood: Heat of combustion ¼ 12.0 kJ/g Heat of gasification ¼ 6.0 kJ/g Effective vaporization temperature ¼ 380 C Radiative loss fraction ¼ 0.35 Conversion constants: 1 ft ¼ 0.308 m 1 gal ¼ 0.003 75 m3 1 lb/ft2 ¼ 4.88 kg/m2 Atmosphere: No wind 25 C Dry air, density ¼ 1.1 kg/m3 (a) 20,000 gal of JP 4 fuel spills entirely over the 87th floor of the building. The fuel experiences a convective heat transfer coefficient of 15 W/m2 K, and a total incident radiative heat flux of 150 kW/m2 due to the flame and compartment effects. Assume a blocking factor of 1. (i) Estimate the burning rate. (ii) Calculate the energy release rate. (iii) Calculate the duration of the JP4 fire. (b) Following the burning off of the jet fuel, the wood-based furnishings burn. The wood fuel loading is distributed such its exposed area to the flames is 0.65 of the floor area. (i) Compute the burning rate of the wood fuel on a floor under identical flame heating conditions as the JP4 in 1. (ii) How long will this fuel burn?

10 Fire Plumes

10.1 Introduction A fire plume is loosely described as a vertically rising column of gases resulting from a flame. The term plume is generally used to describe the noncombustion region, which might dominate the flow away from the combustion source, especially if the source is small. A plume is principally buoyancy driven. This means flow has been induced into the plume due to an increase in temperature or consequent reduction in density. A plume is also likely to be turbulent rather than laminar. Anyone who has observed a rising streak of white smoke from a cigarette left in an ash tray should recall seeing the wavy laminar emitted streak of smoke break-up into a turbulent wider plume in less than 1 foot of height.* Thus, any significant accidental fire will have an associated turbulent plume – even a smoldering fire. Near the fire we will have to deal with a combusting plume, but ‘far’ from the fire it might suffice to ignore the details of combustion entirely. The farfield fire plume acts as if it had been heated by a source of strength Q_ (kW). Since radiated energy from the flame will be lost to the plume, the plume effects will only respond to ð1 Xr ÞQ_ , where Xr is the fraction of radiant energy lost with respect to the actual chemical energy release of the first. We shall see that Xr is an empirical function of fuel and fire geometry. Indeed, much about fire plumes is empirical, meaning that its results are based on experimental correlations. Often the correlations are incomplete with some secondary variables not included since they were not addressed in the experimental study. In other cases, the correlations can be somewhat ambiguous since they represent timeaveraged properties, or length scales (e.g. plume width, flame height), that were not precisely or consistently defined. However, because fire plume results are grounded in measurement – not purely theory – they are practical and invaluable for design and hazard assessment. 3

2

3

ð9:81 m=s Þð0:3 mÞ gz 9 9 Gr ¼ T T1 2 ð5Þ ð20106 m2 =sÞ2 ¼ 3:3 10 . Expect turbulent plumes for Gr > 10 , especially with atmospheric disturbances. *

Fundamentals of Fire Phenomena James G. Quintiere # 2006 John Wiley & Sons, Ltd ISBN: 0-470-09113-4

298

FIRE PLUMES

Some of the foundational studies for turbulent fire plumes include the work of the following investigators: 1. The measurements of Rouse, Yih and Humphries (1952) [1] helped to generalize the temperature and velocity relationships for turbulent plumes from small sources, and established the Gaussian profile approximation as adequate descriptions for normalized vertical velocity (w) and temperature (T), e.g. " # w r 2 ¼ exp wm bðzÞ

ð10:1aÞ

and " # T T1 r 2 ¼ exp bðzÞ Tm T1

ð10:1bÞ

where ðÞm implies maximum or centerline line values and r is the radius. Here, for an axisymmetric plume, b is an effective plume radius that depends on the vertical coordinate z. Such a profile in a scaled variable, ¼ r=bðzÞ, is known as a similarity solution; i.e. the profiles have the same shape at any position z. Near the source where the streamlines will not necessarily follow the vertical direction, a similarity solution will not be possible. The constant in Equation (10.1b) is nearly unity. 2. The theoretical analysis by Morton, Taylor and Turner (1956) [2] established approximate similarity solutions for an idealized point source in a uniform and stably stratified atmosphere. 3. Yokoi (1960) [3], singly produced a ‘small book’ as a report, which carefully investigated point, line and finite heat sources with eventual applications to the hazard from house fire and window flame plumes. 4. Steward (1964) [4] and (1970) [5] presented a rigorous but implicit approximate analysis for both axisymmetric and two-dimensional strip fire plumes respectively, including combustion and variable density effects. All of the solutions are approximate since only a complete solution to the Navier–Stokes unsteady equations with combustion effects would yield fundamental results. This is not currently possible, but will be in the future. These fire plumes have all the scales of turbulent frequencies and specific features due to buoyancy: fine scales – high-frequency turbulence is responsible for the fuel–oxygen mixing and combustion that occurs locally; large turbulent scales – relatively low frequencies – are responsible for the global ‘engulfment’ of air swept into the plume by the large eddy or vortex structures that ‘rollup’ along the outside of a buoyant plume. This is shown schematically in Figure 10.1, and is also shown in a high-speed sequence of photographs by McCaffrey [6] for the luminous flame structure (Figure 10.2). The size of the eddies depend on the diameter of the fire (D) and the energy release rate ðQ_ Þ. The eddy frequency, f, has been shown to

INTRODUCTION

Figure 10.1

Schematic of a turbulent fire plume

correlate with the inverse of a characteristic time, rﬃﬃﬃﬃ g f ¼ 0:48 ðHzÞ D by Pagni [7] in Figure 10.3.

Figure 10.2

299

ð10:2Þ

Intermittency of a buoyant diffusion flame burning on a 0.3 m porous burner. The sequence represents 1.3 s of cine film, showing 3 Hz oscillation (from McCaffrey [6])

300

FIRE PLUMES

Figure 10.3

Eddy shedding frequency of buoyant axisymmetric plumes from Pagni [7]

The mechanism of turbulent mixing which brings air into the buoyant plume is called entrainment. It has been described empirically by relating the momentum of the induced air proportionally to the vertical momentum (mean or centerline), lim ð1 rv2 Þ / r!1

d ðm w2m b2 Þ dz

or dimensionally, the entrainment velocity is traditionally given as rﬃﬃﬃﬃﬃﬃﬃ m wm b lim ðrvÞ ¼ o r!1 1

ð10:3Þ

where o is a constant, the entrainment constant for a Boussinesq or ‘constant density’ plume (i.e. a plume whose density constant is everywhere except in the buoyancy term gz). Many analyses and correlations have ignored the density ratio term in Equation (10.3), treating it as unity. The entrainment constant is not universal, but depends on the mathematical model used. Nominally its value is Oð101 Þ. It must be determined by a modeling strategy together with consistent data. Since all of the theoretical results for plumes rely on experimental data, we must realize that the accuracy of the correlation depends on the accuracy of the data. Measurements of temperature with thermocouples are subject to fouling by soot and radiation error. At flame temperatures, the error can be considerable ð 100 CÞ. Moreover, the fluctuating nature of the properties can lead to fluctuations at a point of 500 CÞ or more. We will see that the maximum time-averaged temperature in turbulent flames can be 800–1200 C, or possibly higher, depending on their size, yet laminar flames can only survive for roughly a flame temperature of >1300 C. Hence, a turbulent flame must be looked on as a collection of highly convoluted, disjointed laminar flame sheets flapping rapidly within a turbulent flow field of large and small eddies. Extinction occurs intermittently due to thermal quenching, flame stretch (shear) and depletion of fuel and oxygen locally. Indeed, the term ‘turbulent diffusion flame’ has lost its favor having been superceded by the term ‘turbulent nonpremixed flames’.

INTRODUCTION

Figure 10.4

301

Flame length intermittency from Zukoski [8]

A buoyancy-dominated turbulent flame has a very complex structure, raising from an oscillating laminar base to a core of luminosity and ending in an intermittent flaming region of broken flame remnants (Figure 10.2). Note that the frequency of flame structure oscillation measured by McCaffrey for a D ¼ 0:3 m burner at 3 Hz is consistent with the vortex shedding frequencies of the Pagni correlation (Equation (10.2)). The intermittency (I ¼ duration of visible flame/flame cycle time) presented by Zukoski [8] in Figure 10.4 shows the flame fluctuations about the mean at I ¼ 0:5. We see about this mean that the flame length, zf , varies by about 20 % for D ¼ 0:1 m. However, the intermittent region expands with D: zf f =2, f ¼ 0:40 zf , 0:60 zf and 0:80 zf for D ¼ 0:10, 0.19 and 0.50 m respectively. Furthermore, the ‘eye-averaged’ zf tends to be about 20 % higher than zf at I ¼ 0:50, according to Zukoski [8]. Measurements of velocity in a plume have more difficulties. Pitot tube measurements rely on temperature and can have ambiguous directional effects. Hot-wire anemometry can be fouled and require buoyancy corrections. Nonintrinsic measurements also offer challenges but have not been sufficiently used in fire measurements. The low-speed flows, especially at the edges of the plume, offer a big measurement challenge. Where a pressure measurement is needed for velocity, appropriate time-averaging of pressure and temperature signals need to be processed. Dynamic pressures can range from 10 to 103 pascals (Pa) in plumes. Realizing that 1 atm is approximately 105 Pa, it is readily approximated that plumes are constant pressure processes. Only vertical static pressure differences are important since they are the source of buoyancy. For large-diameter plumes, especially near their base, it is likely that pressure differences due to the curvature of streamlines can also be responsible for induced flow from the ambient entrainment, but such studies of the base entrainment effect are unknown to this author. Of course, away from the base where streamlines are more parallel to the vertical, pressure effects are nil, and entrainment for the turbulent plume is an engulfment process based on buoyancy and turbulence. Other measurements such as gas species and soot all have importance in fire plumes but will not be discussed here. As we have seen for simple diffusion flames, the mixture fraction plays a role in generalizing these spatial distributions. Thus, if the mixture fraction is determined for the flow field, the prospect of establishing the primary species concentration profiles is possible.

302

FIRE PLUMES

As stated earlier, much has been written about fire plumes. Excellent reviews exist by Zukoski [8], Heskestad [9], Delichatsios [10] and McCaffrey [11]. The student is encouraged to read the literature since many styles of presentation exist, and one style might be more useful than another. We cannot address all here, but will try to provide some theoretical framework for understanding the basis for the many alternative correlations that exist.

10.2 Buoyant Plumes We will derive the governing equations for a buoyant plume with heat added just at its source, approximated here as a point. For a two-dimensional planar plume, this is a line. Either could ideally represent a cigarette tip, electrical resistor, a small fire or the plume far from a big fire where the details of the source are no longer important. We list the following assumptions: 1. There is an idealized point (or line) source with firepower, Q_ , losing radiant fraction, Xr . 2. The plume is Boussinesq, or of constant density 1, except in the body force term. 3. The pressure at any level in the plume is due to an ambient at rest, or ‘hydrostatic’. 4. The ambient is at a uniform temperature, T1 . 5. Properties are assumed uniform across the plume at any elevation, z. This is called a top-hat profile as compared to the more empirically correct Gaussian profile given in Equation (10.1). 6. All properties are constant, and cpi ¼ cp .

10.2.1

Governing equations

Later we shall include combustion and flame radiation effects, but we will still maintain all of assumptions 2 to 5 above. The top-hat profile and Boussinesq assumptions serve only to simplify our mathematics, while retaining the basic physics of the problem. However, since the theory can only be taken so far before experimental data must be relied on for its missing pieces, the degree of these simplifications should not reduce the generality of the results. We shall use the following conservation equations in control volume form for a fixed CV and for steady state conditions: Conservation of mass (Equation (3.15)) j X

m_ j ¼ 0

ð10:4Þ

net out

Conservation of vertical ðzÞ Momentum (Equation (3.23)) þz "

X

Fz ¼

j X net out

m_ j wj

ð10:5Þ

BUOYANT PLUMES

Figure 10.5

303

(a) Schematic of a point source plume and (b) surface focus

Conservation of energy (Equation (3.48)) cp

j X

m_ j Tj ¼ Q_ q_ loss

ð10:6Þ

net out

We will apply these to a small control volume of radius b and height dz, as shown in Figure 10.5, and we will employ the Taylor expansion theorem to represent properties at z þ dz to those at z: Fðz þ dzÞ ¼ FðzÞ þ

dF dz þ OðdzÞ2 dz

ð10:7Þ

304

FIRE PLUMES

Using the entrainment relationship (Equation (10.3)), in terms of the top-hat profile, ve ¼ o w

ð10:8Þ

The conservation of mass can then be written as m_ ðz þ dzÞ m_ ðzÞ 1 ve 2 b ¼ 0

ð10:9Þ

The mass flow rate in terms of the Boussinesq top-hat assumption is given by m_ ¼ 1 w b2

ð10:10Þ

Therefore, the conservation of mass using Equations (10.7) to (10.10) with the operation lim

1

dz!0 dz

gives d ð1 w b2 Þ ¼ 1 o w b dz or d ðwb2 Þ ¼ 2o wb dz

ð10:11Þ

The conservation of momentum in the z direction with positive forces in the þz directions can be written as p1 2 b db þ ðp1 b2 Þz ðp1 b2 Þzþdz g b2 dz s 2 b dz ¼ ðwm_ Þzþdz ðwm_ Þz ð10:12Þ The pressure forces are derived in Figure 10.5(b), taking into account the pressure at the perimeter region 2 b ds, where b is a mean between z and z þ dz. Likewise, so is p1 . The vertical component of the shear force also follows from similar arguments. However, at the edge of the plume we have a static fluid (approximately) or dw=dr ¼ 0 for the actual profile. Since

s /

dw dr

realizing that we are considering turbulent time-averaged quantities, then s ¼ 0 at r ¼ b. Therefore, in the limit as dz ! 0, p1 2 b

db d d ðp1 b2 Þ g b2 ¼ ð1 w2 b2 Þ dz dz dz

ð10:13Þ

BUOYANT PLUMES

305

However, in the static ambient, dp1 ¼ 1 g dz

ð10:14Þ

d ð1 w2 b2 Þ ¼ ð1 Þg b2 dz

ð10:15Þ

Combining gives

Employing the perfect gas equation of state, T ¼ 1 T1

ð10:16Þ

To reduce non-linear effects, we make another assumption: 7. The change in density or temperature is small. (This completes the full Boussinesq model.) We shall retain this assumption even for the high-temperature combustion case, realizing that it is quite severe. However, its neglect will undoubtedly be compensated through ultimate experimental correlations. With assumption 7 we approximate 1 T1 T T1 T T1 T1 ¼ ¼ ¼1 1 T T T1 T 1 T T1 for T=T1 ! 1 1 T1

ð10:17Þ

Therefore, Equation (10.15) becomes d 2 2 ðw b Þ ¼ dz

T T1 gb2 T1

ð10:18Þ

The conservation of energy for the differential control volume in Figure 10.5 becomes, following Equation (10.6), cp ½ðm_ TÞzþdz ðm_ TÞz 1 ve cp T1 2 b dz

ð10:19Þ

The heat added is zero because we have no combustion in the CV, and all of the heat is ideally added at the origin, z ¼ 0. Operating as before, we obtain cp

d ð1 w b2 ÞT 1 ve cp T1 2 b ¼ 0 dz

From Equation (10.11), it can be shown that cp

d d ð1 w b2 TÞ cp ð1 w b2 T1 Þ ¼ 0 dz dz

ð10:20Þ

306

FIRE PLUMES

or 1 cp

d ½w b2 ðT T1 Þ ¼ 0 dz

ð10:21Þ

This can be integrated, and using a control volume enclosing z ¼ 0 yields 1 cp w b2 ðT T1 Þ ¼ Q_ ð1 Xr Þ where Q_ is the origin term representing the power of the source. Hence, wb2

T T1 Q_ ð1 Xr Þ ¼ 1 cp T1 T1

ð10:22Þ

We now have three equations, (10.11), (10.18) and (10.22), to solve for w, b, and ¼ ðT T1 Þ=T1 . The initial conditions for this ideal plume are selected as (at z ¼ 0) b¼0

ða point sourceÞ

and w2 b ¼ 0

ðno momentumÞ

If we had significant momentum or mass flow rate at the origin, the plume would resort to a jet and the initial momentum would control the entrainment as described by Equation (10.3). This jet behavior will have consequences for the behavior of flame height compared to the flame height from natural fires with negligible initial momentum.

10.2.2

Plume characteristic scales

It is very useful to minimize significant variables by expressing the equations in dimensionless terms. To do this we need a characteristic length to normalize z. If we had a finite source diameter, D, it would be a natural selection; however, it does not exist in the point source problem. We can determine this natural length scale, zc , by exploring the equations dimensionally. We equate dimensions. From Equation (10.22), D

wc z2c ¼

Q_ 1 cp T1

or D

wc ¼

Q_ 1 cp T1 z2c

Substituting into Equation (10.18) gives D

w2c zc ¼ gz2c

BUOYANT PLUMES

307

or

2 Q_ D z1 c ¼ g 1 cp T1 z2c

giving

2=5 Q_ pﬃﬃﬃ 1 cp T1 g

zc ¼

ð10:23Þ

as the natural length scale. This length scale is an important variable controlling the size of the large eddy structures and the height of the flame, in addition to the effect by D for a finite fire. We might even speculate that the dimensionless flame height should be expressed as

z zf c ¼f D D Incidentally, if we were examining a line source in which the source strength would need to be expressed as Q_ 0 , i.e. energy release rate per unit length of the linear source, Equation (10.23) could be reexamined dimensionally as D

zc ¼

2=5 Q_ =L pﬃﬃﬃ 1 cp T1 g=zc

or zc;line ¼

2=3 Q_ 0 pﬃﬃﬃ 1 cp T1 g

ð10:24Þ

This is the characteristic length scale for a line plume. To complete the point source normalization, we determine, in general, a characteristic velocity as wc ¼

pﬃﬃﬃﬃﬃﬃ gzc

ð10:25aÞ

or specifically for a point source,

wc ¼ g1=2

where Q_ is the firepower in this scaling.

1=5 Q_ pﬃﬃﬃ 1 cp T1 g

ð10:25bÞ

308

FIRE PLUMES

10.2.3

Solutions

Solutions for a Gaussian profile, with Equation (10.1) for an axisymmetric point source, and for a line source w y 2 ¼ exp wm b

ð10:26aÞ

T T1 y 2 ¼ exp b Tm T1

ð10:26bÞ

and

with dimensions as shown in Figure 10.6, are tabulated in Table 10.1 [12]. The dimensionless variables are ¼ ðTm T1 Þ=T1 W ¼ wm =wc B ¼ b=zc

¼ z=zc

ð10:27Þ

The solutions in Table 10.1 were fitted to data by Yokoi [3] from small fire sources with an estimation of Xr , the radiation fraction, for his fuel sources. Fits by Zukoski [8], without accounting for radiation losses ðXr ¼ 1Þ, give values of o ¼ 0:11 and ¼ 0:91 for the axisymmetric source. Yuan and Cox [13] have shown that o and can range from 0.12 to 0.16 and from 0.64 to 1.5 respectively, based on a review of data from many studies. Figures 10.7 and 10.8 show the similarity results and centerline characteristics respectively by Yokoi [3] for the point and line sources. It is interesting to note some features of these ideal cases since they match the far-field data of large fire plumes. The velocity and temperature at the centerline of a point plume decreases with height and both are singular at the origin. Only the temperature has this behavior in the line plume, with the velocity staying constant along the centerline. A

Figure 10.6

Finite sources

BUOYANT PLUMES Far-field Gaussian plume correlations [12]

Table 10.1 Dimensionless variable

Axisymmetric (Xr ¼ 0:20 for data) Cl Cl ¼ 65 ; 0:118

B W

Cv 1=3

1=3 25 þ 1 2 ð1 Xr Þ 24

Cv ¼

309

Infinite line (Xr ¼ 0:30 for data) Cl Cl ¼ p2ﬃﬃ ; 0:103 Cv 0 " # 1=6 þ1 Cv ¼ 1=3 ð1 Xr Þ1=3 2

Cv ¼ 4:17ð1 Xr Þ1=3

Cv ¼ 2:25ð1 Xr Þ1=3

CT 5=3 " # 2=3 2 25 ð þ 1Þ2=3 2=3 CT ¼ ð1 X Þ r 3 24 1=3 4=3

CT 1 " # ð þ 1Þ1=3 2=3 2=3 ð1 X Þ CT ¼ r 25=6 1=6

CT ¼ 10:58ð1 Xr Þ1=3

CT ¼ 3:30ð1 Xr Þ2=3 qﬃﬃ 2 1:54 ;

2

W F

3=2 ;

1:64

0.098 0.913

Figure 10.7

0.091 0.845

Experimental results of Yokoi (BRI, Japan) for point and line sources [3]

310

FIRE PLUMES

Figure 10.8

Yokoi data for time-averaged axial values of velocity and temperature rise [3]

local Froude number (Fr) is invariant and nearly the same for both plumes. This local Fr number is represented as kinetic energy potential energy w2m =g zc w2m W2 ¼ ¼ 1:6 0:05 ¼ ½ðTm T1 Þ=T1 ðz=zc Þ g½ðTm T1 Þ=T1 z Fr

ð10:28Þ

as seen from Table 10.1. This shows that velocity is directly related to buoyancy. Note that no effect of viscosity has come into the analysis for the free plume. Indeed, an application of Bernoulli’s equation to the center streamline would give a value of Fr ¼ 2:0. For wall plumes, or ceiling–plume interactions, we might expect to see effects of Gr ¼ ðgz3 = 2 Þ½ðTm T1 Þ=T1 . For laminar plumes, viscous effects must be present since viscosity is the only mechanism allowing ambient air to be ‘entrained’ into the plume flow.

COMBUSTING PLUMES

311

10.3 Combusting Plumes We will now examine the combustion aspect of plumes and attempt to demonstrate more realistic results. The preceding approach provides a guide for how we might proceed. Since the complexities of combustion are too great to consider in any fundamental way, we simply adopt the same ‘point source’ approach with the additional assumption: 8. Combustion occurs uniformly throughout the plume so long as there is fuel present. The equations become the same as before, except with a net effective heat addition per unit volume as ð1 Xr ÞQ_ 000 within the differential element of Figure 10.5. From Equations (10.11), (10.18) and (10.21), the conservation of mass, momentum and energy become d ðwb2 Þ ¼ 2o wb dz d 2 2 T T1 ðw b Þ ¼ gb2 dz T1

ð10:29aÞ ð10:29bÞ

and 1 cp

d ½w b2 ðT T1 Þ ¼ ð1 Xr ÞQ_ 000 b2 dz

ð10:29cÞ

for the top-hat axisymmetric case. The maximum rate of fuel that can burn in the control volume dz in Figure 10.5(a) is that which reacts completely with the entrained oxygen or with a known stoichiometric ratio of oxygen to fuel, r. Thus, we can write 1 ðo wÞð2 b dzÞðhc YO2 ;1 Þ Q_ 000 ¼ ð b2 dzÞðrÞ

ð10:30Þ

Substituting in Equation (10.29c) gives 1 cp

d ð1 Xr Þ2o 1 wbhc YO2 ;1 ½wb2 ðT T1 Þ ¼ dz r

ð10:31Þ

Using Equation (10.29a), this becomes d hc YO2 ;1 d ½wb2 ðT T1 Þ ¼ ð1 Xr Þ ðwb2 Þ dz dz cp r

ð10:32Þ

which integrates to give, with b ¼ 0 at z ¼ 0, T T1 ¼ ð1 Xr ÞYO2 ;1

hc rcp

ð10:33Þ

However, it is not likely that all of the air entrained will react with the fuel. The turbulent structure of the flame does not allow the fuel and oxygen to instantaneously mix. There is a delay, sometimes described as ‘unmixedness’. Hence, some amount of

312

FIRE PLUMES

oxygen (or air) more than stoichiometric will be entrained but not burned. Call this excess entrainment fraction mass of air entrained ð10:34Þ n mass of air reacted Equation (10.33) should be modified to give T T1 ¼

ð1 Xr ÞYO2 ;1 hc ncp r

ð10:35Þ

a lower temperature in the flame due to dilution by the excess air. This factor n has been found to be about 10, and may be a somewhat universal constant for a class of fire plumes. The conditions for this constancy have not been established. The main point of this result is that the time and spatial average flame temperature is a constant that depends principally on Xr alone since YO2 ;1 hc =r 2:9 kJ=g air for a wide range of fuels. Indeed, the flame temperature is independent of the fuel except for its radiative properties, and those flames that radiate very little have a higher temperature. This result applies to all fire plumes configurations, except that n will depend on the specific flow conditions. In dimensionless terms, for buoyancy-dominated turbulent fires, we would expect Tm T1 ¼ f ð =nÞ T1

ð10:36Þ

where

ð1 Xr ÞYO2 ;1 hc cp T1 r

is a dimensionless parameter representing absorbed chemical to sensible enthalpy. Table 10.2 gives correlation results based on Gaussian profiles with selected as 1 for the axisymmetric and line-fire plumes [12]. It is indeed remarkable that the ‘local’ Table 10.2 Dimensionless variable

Axisymmetric (Xr ¼ 0:20 for data)

Infinite line (Xr ¼ 0:30 for data)

Cl 4 Cl ¼ ; 0:179 5 Cv 1=2 pﬃﬃﬃﬃ Cv ¼ 2:02 or 0:720

Cl 4 Cl ¼ pﬃﬃﬃ ; 0:444 3 Cv 1=2 pﬃﬃﬃﬃ Cv ¼ 2:3 or 0:877

CT 0

CT 0

CT ¼ 2:73 or 0:347

CT ¼ 3:1 or 0:450

W2

1:49

1:70

0.22 1

0.590 1

B

W

a

Combusting region flame correlations [12]

pﬃﬃﬃﬃ from McCaffrey data [6], but 0.50 is recommended.

FINITE REAL FIRE EFFECTS

313

Froude number W 2 =ð Þ is nearly the same for the near and far field plumes, 1:60 0:1. The entrainment coefficients are much larger, which probably includes pressure effects near the base for the axisymmetric fires and ‘tornado-flame filament’ effects for the line fire, which are actually three dimensional. It should be realized that the data corresponding to these correlations contain results for finite fires: D > 0, not idealized sources. The correlations in Tables 10.1 and 10.2 are one set of formulas; others exist with equal validity.

10.4 Finite Real Fire Effects Let us examine some alternative results and consider the effects of the base fire dimension D. Measurements are given by Yokoi [3] for fire sources of Figure 10.9 and 20 cm radius ðro ¼ D=2Þ. The temperature profiles T T1 Tm T1

versus

z ro

and

r ro

are plotted in Figure 10.9. Similarity is preserved, although the Gaussian profile is not maintained in the fire zone. Also the plume radius grows more slowly in the combustion zone. This slower growth is inconsistent with theory, as presented in Table 10.2, where C‘ values are constant since the simple entrainment theory is not adequate in the combustion zone.

10.4.1

Turbulent axial flame temperatures

McCaffrey [6] gives results for axisymmetric fires with a square burner of side 0.3 m using methane. The results in Figure 10.10 clearly delineates three regions: (1) the

Figure 10.9

Dimensionless temperature profiles, ðT T1 Þ=ðTm T1 Þ, for finite fire sources (from Yokoi [3])

314

FIRE PLUMES

Figure 10.10

Centerline fire plume (a) temperature rise and (b) velocity (from McCaffrey [6])

continuous luminous flame, (2) the oscillating flame zone and (3) the noncombusting plume region. Similar results are displayed by Quintiere and Grove [12] in Figure 10.11 for infinite line fires of widths D ranging from about 0.1 mm to 20 cm and Q_ 0 from about 102 to 350 kW/m. Note that the maximum flame temperature appears to be 800–900 C for these scale fires. For larger scale fires, this flame temperature can increase with diameter; e.g. for D 2 m,

FINITE REAL FIRE EFFECTS

Figure 10.11

315

Dimensionless centerline fire plume (a) temperature rise and (b) velocity for infinite line fires of width D [12]

900 C; 6 m, 1000 C; 15 m, 1100 C; and 30 m, 1200 C [14]. The explanation is provided by Koseki [15] (Figure 10.12), showing how Xr decreases for large-diameter fires as eddies of black soot can obscure the flame. The eddy size or soot path length increases as the fire diameter increases, causing the transmittance of the external eddies to decrease and block radiation from leaving the flame. From Table 10.2,

316

FIRE PLUMES

Figure 10.12 Relationship between the radiative fraction and pan diameter for various fuel fires (from Koseki [15])

the flame temperature can be found from Tf T1 YO2 ;1 hc ¼ CT;f ð1 Xr Þ T1 rcp T1

ð10:37Þ

Table 10.3 gives values of CT;f for carefully measured temperatures (50 m Pt–Rh [16] and 1 mm [17] thermocouples); CT;f ¼ 0:50 is recommended for cp ¼ 1 kJ=kg K. This holds in the flame core region where combustion always exists, z < zf . From Koseki’s Figure 10.12, ðTf T1 ÞD¼30 m 1 Xr ð30 mÞ 1 0:06 ¼ 1:7 ¼ ðTf T1 ÞD¼0:3 m 1 Xr ð0:3 mÞ 1 0:45 This is consistent with the temperature measurements reported by Baum and McCaffrey [14], ðTf T1 ÞD¼30 m 1200 C ¼ 1:5 800 C ðTf T1 ÞD¼0:3 m Application of Equation (10.37) gives a maximum value of the flame temperature rise corresponding to Xr ¼ 0 as Tf T1 1450 C Table 10.3 Maximum time averaged centerline temperature ( C) Fuel 1260 1100 930 990

Natural gas Natural gas Heptane Heptane

Maximum measured centerline flame temperature hc/s (kJ/g air) D (m) Q_ (kW) 2.91 2.91 2.96 3.07

0.3 0.3 1.74 1.74

17.9 70.1 7713 973

Xr 0.15 0.25 0.43 0.18

Source

CT;f

Smith and Cox [17] 0.50 Smith and Cox [17] 0.48 Kung and Stavrianidis [18] 0.52 Kung and Stavrianidis [18] 0.59

FINITE REAL FIRE EFFECTS

Figure 10.13

317

Virtual origin

Although such a value has not been reported, sufficient care has not generally been taken to obtain accurate measurements in large fires.

10.4.2

Plume temperatures

It is obvious that idealized analyses of point or line source models for fire plumes have limitations. Only at large enough distances from the fire source should they apply. One approach to bring these idealized models into conformance with finite diameter fire effects has been to modify the analysis with a virtual origin, zo . For a finite source, a simple geometric adaption might be to locate an effective point source of energy release at a distance zo below the actual source of diameter, D. This is illustrated in Figure 10.13 where the location of zo is selected by making the plume width to coincide with D at z ¼ 0. Various criteria have been used to determine zo from fire plume data. From the farfield solution for temperature modified for the virtual (dimensionless) origin below the actual source, we can write from Table 10.1 in general T T1 ¼ CT ð1 Xr Þ2=3 ð þ o Þ5=3 T1

ð10:38Þ

pﬃﬃﬃ where o ¼ zo =½Q_ =ð1 cp T1 gÞ2=5 . This equation can be rearranged so that a new temperature variable can be plotted as a linear function of z: h

z ¼ zo þ

i2=5 Q_ ð1Xr Þ pﬃﬃ 1 cp T 1 g ð10:58Þ3=5

3=5 TT1 T1

ð10:39Þ

Such a plot of data is illustrated by Cox and Chitty [18] in Figure 10.14. The regression line favors the data in the intermittent flame region and in the plume immediately above the flame. For these data the virtual origin is 0.33 m below the burner. In the continuous

318

FIRE PLUMES

Figure 10.14

Axial temperature data plotted to reveal the virtual origin (from Cox and Chitty [18])

flame zone below z ¼ 0:4 m, the flame temperature is constant, as expected. For data above z ¼ 1:2 m, a linear fit for those data could yield a zo close to zero – more consistent for far-field data. The level of uncertainty of such deductions for the correction factor zo should be apparent. A result for zo must be used in the context of the specific data correlation from which it was obtained. For example, Heskestad [19] derives a correlation for the virtual origin as zo 0:083 Q_ 2=5 ¼ 1:02 D D

ð10:40aÞ

where zo and D are in m and Q_ is in kW. In dimensionless form this is zo 2=5 ¼ 1:02 1:38 QD D

ð10:40bÞ

Q_ pﬃﬃﬃ 1 cp T1 gD5=2

ð10:41Þ

where QD

Heskestad [19] determined this result for use with the centerline far-field plume temperature correlation for z zf , given as T T1 ¼ 9:1ð1 Xr Þ2=3 ð þ o Þ5=3 T1

ð10:42Þ

which represents the best version of Equation (10.38). Heskestad limits the use of this equation to T T1 500 C, or above the mean flame height. The data examined in

FINITE REAL FIRE EFFECTS

319

Table 10.4 Axisymmetric fire plume data for varying fuels and diameter taken from References [16], [17] and [19] Fuel Silicone fluid Silicone fluid Natural gas, sq.a Methanol Methanol Methanol Natural gas, sq.a Natural gas, sq.a Hydrocarbon fluid Natural gas, sq.a Hydrocarbon fluid Natural Gas, sq.a Heptane Heptane Heptane a

QD

D (m)

Q (kW)

Xr

zo =D

zf =D

0.038 0.049 0.19 0.19 0.22 0.26 0.29 0.44 0.49 0.60 0.61 0.77 1.75 1.93 1.93

2.44 1.74 0.34 2.44 1.74 1.22 0.34 0.34 1.74 0.34 1.22 0.34 1.74 1.22 1.22

406 211 14.4 1963 973 467 21.7 33.0 2151 44.9 1101 57.5 7713 3569 3517

0.19 0.19 — 0.24 0.18 0.12 — — 0.28 — 0.28 — 0.43 0.41 0.36

0.30 0.50 0.29 0.14 0.16 0.36 0.18 0.06 0.43 0.09 0.10 0.09 0.71 0.43 0.43

0.14 0.11 — 0.98 0.98 0.74 — — 2.1 — 1.8 — 3.2 3.2 3.2

sq. implies a square rather than a circular burner was used.

deriving Equations (10.40) and (10.42) are displayed in Table 10.4 and represent a wide range of conditions. The parameter QD is a dimensionless energy release rate of the fire. It represents the chemical energy release rate divided by an effective convective energy transport flow rate due to buoyant flow associated with the length scale D. (It was popularized by Zukoski [8] in his research and might be aptly designated the Zukoski number, Zu Q .) For natural fires, the energy release rate for pool fires of nominal diameter, D, depends on the heat transfer from the flames in which the energy release rate for a fixed diameter ðDÞ can be widely varied by controlling the fuel supply rate. Therefore burner flames are somewhat artificial natural fires. Hasemi and Tokunaga [20] plots a range of natural fires in terms of fuel type, nominal diameter D and QD , as shown in Figure 10.15. The results show that the range for QD for natural fires is roughly 0.1 to 10.

10.4.3

Entrainment rate

Equation (10.10) gives the mass flow rate in the plume. This is exactly equal to the entrainment rate if we neglect the mass flow rate of the fuel. The latter is small, especially as z increases. For an idealized point source, the entrainment rate consistent with the farfield results of Table 10.1 is m_ e ¼ Ce 5=6 pﬃﬃﬃ 1 g z5=2

ð10:43aÞ

or in terms of QD , far-field entrainment is found from 5=6 m_ e 1=3 z pﬃﬃﬃ 5=2 ¼ Ce QD D 1 g z

ð10:43bÞ

320

FIRE PLUMES

Figure 10.15

Natural fires in terms of QD by Hasemi and Tokunaga [20]

The coefficient Ce depends on Xr and is 0.17 for Xr ¼ 0:2 for the empirical constants of Table 10.1; Zukoski [8] reports 0.21 and Ricou and Spalding [21] report 0.18. For the far-field, valid above the flame height, Heskestad [22] developed the empirical equation: n o m_ e ðkg=sÞ ¼ 0:071½ð1 Xr ÞQ_ 1=3 ðz þ zo Þ5=3 1 þ 0:026½ð1 Xr ÞQ_ 2=3 ðz þ zo Þ5=3 ð10:44Þ where Q_ is in kW and z and zo are in m; zo is given by Equation (10.40). While Eq. (10.43) is independent of D, Equation (10.44) attempts to correct for the finite diameter of the fire through the virtual origin term. The development of an entrainment correlation for the flame region is much more complicated. Data are limited, and no single correlation is generally accepted. Delichatsios [10] shows that the flame entrainment data for burner fires of 0.1–0.5 m in diameter can be correlated over three regions by power laws in z=D depending on whether the flame is short or tall. We shall develop an equivalent result continuous over the three regions. We know the plume width is a linear function of z by Table 10.2, as well as suggested by the data of Yokoi in Figure 10.9. Although the visible flame appears to ‘neck’ inward and must close at its tip, the edge of the turbulent flame plume

FINITE REAL FIRE EFFECTS

321

generally expands from its base. Therefore, for a fire of finite diameter, it is reasonable to represent the plume radius, b, as b¼

D þ C‘ z 2

ð10:45Þ

where C‘ is an empirical constant. Due to the constancy of W2 = , w is proportional to z1=2 in the flame zone (see Table 10.2 and Figure 10.10(b)). Hence, substituting these forms for b and w into Equation (10.10) gives 2 1=2 D þ C‘ z ð10:46aÞ m_ e ¼ 1 Cv z 2 or in dimensionless form, entrainment in the flame is found from

z 1=2 h

z i2 m_ e 1 þ 2C‘ pﬃﬃﬃ 5=2 ¼ Ce D D 1 g D

ð10:46bÞ

where C‘ ¼ 0:179 and Ce ¼ 0:0571=2 , approximately. These coefficients have been found to fit the data for burners of 0.19 and 0.5 m as given by Zukoski [8]. The results are plotted in Figure 10.16. Since velocity depends on plume temperature and temperature depends on radiation loss, it should be expected that Ce depends on Xr . However, it is not entirely obvious, nor is it generally accepted, that this flame entrainment rate is independent of Q_ . This fact was demonstrated some time ago by Thomas, Webster and Raftery [23] and is due to the constancy of the flame temperature regardless of Q_ . Essentially, the perimeter flame area is significant for entrainment.

Figure 10.16 Flame entrainment rate by data from Zukoski [8] and the correlation of Quintiere and Grove [12]

322

FIRE PLUMES

10.4.4

Flame height

The flame height is intimately related to the entrainment rate. Indeed, one is dependent on the other. For a turbulent flame that can entrain n times the air needed for combustion (Equation (10.34)), and r, the mass stoichiometric oxygen to fuel ratio, the mass rate of fuel reacted over the flame length, zf , is m_ F ¼

m_ e YO2 ;1 nr

ð10:47Þ

where m_ e is taken from Equation (10.46) for z ¼ zf and YO2 ;1 is the oxygen mass fraction of the ambient. By multiplying by hc, the heat of combustion (Equation (10.47)) can be put into dimensionless form as YO2 ;1 hc m_ e pﬃﬃﬃ rcp T1 1 g D5=2 ð10:48Þ QD ¼ n Substitution of Equation (10.46b) with z ¼ zf gives a correlation for flame height. The value n is found by a best fit of data in air ðYO2 ;1 ¼ 0:233Þ as 9.6. The implicit equation for zf in air is QD

z i2 3=2 zf 1=2 h f ¼ 0:005 90 1 þ 0:357 1 Xr D D

ð10:49aÞ

In air for zf =D > 0:1, an approximation for the function in zf =D gives 2=5

zf QD 16:8 1:67 D ½ðhc =sÞ=ðcp T1 Þ3=5 ð1 Xr Þ1=5

ð10:49bÞ

and for zf =D < 0:1, emphasizing the half-power, zf Q2 D 2:87 104 D ½ðhc =sÞ=ðcp T1 Þ3 ð1 Xr Þ

ð10:49cÞ

Here s is the stoichiometric air to fuel ratio. An alternative, explicit equation for zf given by Heskestad [24], based on the virtual origin correction (see Equation (10.40)), is given as 3=5 zf 2=5 scp T1 ¼ 15:6 QD 1:02 D hc

ð10:50aÞ

or zf ðmÞ ¼ 0:23 Q_ 2=5 1:02 DðmÞ;

Q_

in

kW

ð10:50bÞ

FINITE REAL FIRE EFFECTS

323

Figure 10.17 Axisymmetric flame height in terms of QD [12]

The equations are compared in Figure 10.17 along with a region of data taken from Zukoski [8]. Additional data presented by Zukoski [25] for the small QD range are presented in Figure 10.18. The straight lines show that the dependencies, consistent with 2=5 Equation (10.49), are for small QD , zf =D Q2 D , and for large QD , zf =D QD . However, this large QD behavior only applies as long as initial fuel momentum effects pﬃﬃﬃﬃﬃﬃ are negligible compared to the buoyancy velocity, i.e. wF (fuel velocity) g z. Once the fuel injection velocity is large, the process becomes a jet flame.

10.4.5

Jet flames

Hawthorne, Weddell and Hottel [26] showed that the height of a turbulent jet flame is approximately constant for a given fuel and nozzle diameter. If it is truly momentum dominated, the flame length will not change with increasing fuel flow rate. This can approximately be seen by using similar reasoning as in Section 10.4.4, but realizing that in the flame region the fuel nozzle velocity, wF , dominates. From Equations (10.10) and (10.45), with w ¼ wF , we write for the entrainment of the jet flame: 2 D þ C‘; j z ð10:51Þ m_ e 1 wF 2 Turbulent jet flames have zf D, so we can neglect D in Equation (10.51). The fuel mass flow rate is given by m_ F ¼ F D2 wF ð10:52Þ 4

324

FIRE PLUMES

Figure 10.18

Flame lengths for small Q from Zukoski [25]

where F is the fuel density. Substituting Equations (10.51) and (10.52) into Equation (10.47) gives C‘;2 j ðzf =DÞ2 1 m_ F m_ F YO2 ;1 F nj r or 1=2 pﬃﬃﬃﬃ nj F r zf D 2C‘; j 1 YO2 ;1

ð10:53Þ

where nj and C‘;j are constants. Hence, turbulent jet flame height depends on nozzle diameter, fuel density and stoichiometry. McCaffrey [6] shows that this behavior occurs between QD of 105 to 106 . Note QD ¼

F wF hc pﬃﬃﬃﬃﬃﬃ 4 1 gD cp T1

showing QD is related to a fuel Froude number w2F =ðgDÞ.

ð10:54Þ

FINITE REAL FIRE EFFECTS

10.4.6

325

Flame heights for other geometries

It is important to be able to determine the turbulent flame heights for various burning configurations and geometrical constraints. Since flame length is inversely related to air or oxygen entrainment, constraints that limit entrainment, such as a fire against a wall or in a corner, will result in taller flames for the same Q_ . For example, Hasemi and Tokunaga [20] empirically find, for a square fire of side D placed in a right-angle corner, 2=3

zf =D ¼ Cf QD ;

Q_

in

kW

ð10:55Þ

where Cf ¼ 4:3 for the flame tip and Cf ¼ 3:0 for the continuous flame height. The interaction of axisymmetric flames with a ceiling suggests that the radial flame extension is approximately Rf Cf ðzf;o HÞ

ð10:56Þ

where Cf can range from 0.5 to 1, zf;o is the height of the free flame with no ceiling and H is the height of the fuel source to the ceiling [27,28]. For narrow line fires of D=L 0:1 (see Figure 10.6), it can be shown that

z zf c;line ¼ function D D following Equation (10.24). An empirical fit to the collection of data shown in Figure 10.19 gives [12]

z i Q_ 0 3=2 zf 1=2 h f 1 þ 0:888 ð10:57Þ pﬃﬃﬃ 3=2 ¼ 0:005 90 1 Xr D D 1 cp T1 g D

Figure 10.19 Flame height for a line fire of width D [12]

326

FIRE PLUMES

Some data for small D (1.5 cm) and zf =D < 10 have been observed to be laminar; other data for zf =D fall below the correlation, suggesting inconsistent definitions for flame height. Equation (10.57) is the counterpart to Equation (10.49) for the axisymmetric case, and was developed in a similar manner. Wall fires are an extension of the free plume line and axisymmetric fires. Based on the data of Ahmad and Faeth [29], for turbulent wall fires, it has been shown that zf ¼

1:02

Q_ 0 pﬃﬃﬃ 1 cp T1 g

YO2 ;1 ð1 Xr Þ1=3

2=3 ð10:58Þ

For a square fire of side D against a wall Hasemi and Tokunaga [20] find that zf 2=3 ¼ Cf QD D

ð10:59Þ

where Cf ¼ 3:5 for the flame tip and Cf ¼ 2:2 for the continuous flame.

10.5 Transient Aspects of Fire Plumes For simplicity let us consider an idealized point (or line) transient plume. Then by an extension of Equation (10.27), it can be shown that the plume dimensionless variables would depend on the axial position and time as ( ) z t f; W; Bg as functions of ; pﬃﬃﬃﬃﬃﬃﬃﬃﬃ zc zc =g One approach is to approximate the unsteady solution by a quasi-steady solution where Q_ ðtÞ is a function of time and time is adjusted. This can be approximated by determining the transport time ðto Þ from the origin to position z by to ¼

Z

z o

dz w

ð10:60Þ

where w is taken as the steady state solution with Q_ ¼ Q_ ðtÞ. Then the unsteady solution can be approximated by adjusting t to t þ to at z, e.g. z ðt þ to Þ ; pﬃﬃﬃﬃﬃﬃﬃﬃﬃ zc zc =g

!

steady

z _ ; QðtÞ zc

ð10:61Þ

More specific results are beyond the scope of our limited presentation for plumes. However, we will examine some gross features of transient plumes; namely (a) the rise of a starting plume and (b) the dynamics of a ‘fire ball’ due to the sudden release of a finite burst of gaseous fuel. Again, our philosophy here is not to develop exact solutions, but to represent the relevant physics through approximate analyses. In this way, experimental correlations for the phenomena can be better appreciated.

327

TRANSIENT ASPECTS OF FIRE PLUMES

Figure 10.20

10.5.1

Starting plume

Starting plume

We consider a point source axisymmetric example as illustrated in Figure 10.20. We shall examine the rise of the plume, zp , as a function of time. As the hot gases rise due to the source Q_ , initiated at time t ¼ 0, the gases at the ‘front’ or ‘cap’ encounter cooler ambient air. The hot gases in the cap, impeded by the air, form a recirculating zone, as illustrated in Figure 10.20. Entrainment of air occurs over the vertical plume column and the cap. The warmed, entrained air forms the gases in the plume. The height of the rising plume can be determined from a conservation of mass for a control volume, enclosing the plume as it rises. From Equation (3.15), it follows that dmp ¼ m_ e þ m_ F dt

ð10:62Þ

where mp ¼ mass of the plume m_ e ¼ rate of air entrained m_ F ¼ rate of fuel mass supplied We make the following assumptions in order to estimate each term: 1. The rate of fuel supply is small compared to the rate of air entrained, m_ F 0. 2. The plume properties can be treated as quasi-steady. This is equivalent to applying Equation (10.60) with to small or negligible compared to t. Then at an instant of time the steady state solutions apply. 3. In keeping with assumption 2, the plume cap is ignored, and the top of the rising plume is treated as a steady state plume truncated at zp . 4. The rising plume has a uniform density, 1 , equal to the ambient. 5. The rising plume can be treated as an inverted cone of radius, b, as given in Table 10.1, b ¼ 65 o z.

328

FIRE PLUMES

As a consequence, the mass of the plume can be expressed as 2 6 o zp zp mp ¼ 1 3 5

ð10:63Þ

Substituting into Equation (10.62), along with Equation (10.43) for m_ e, gives 1

13 2 d 3 pﬃﬃﬃ 5=3 o ðzp Þ ¼ 1 gCe z5=6 c zp 25 dt

ð10:64Þ

where zc is the characteristic plume length scale, given in Equation (10.23). Rearranging, and realizing that o and Ce are particular constants associated with the entrainment characteristics of the rising plume, we write dz3p dt

pﬃﬃﬃ 5=3 ¼ Cp0 gz5=6 c zp

ð10:65Þ

where Cp0 is an empirical constant. Integrating from t ¼ 0, zp ¼ 0, we obtain Z zp 0 pﬃﬃﬃ 5=6 z1=3 p dzp ¼ Cp gzc t 0

or 3 4=3 pﬃﬃﬃ z ¼ Cp0 gz5=6 c t 4 p

ð10:66Þ

In dimensionless form, with a new empirical constant, Cp , pﬃﬃﬃ 4=3 gt zp ¼ Cp pﬃﬃﬃﬃ zc zc

ð10:67Þ

Zukoski [8] claims that Cp can range from 3 to 6, and suggests a value from experiments by Turner [30] of 3.5. Recent results by Tanaka, Fujita and Yamaguchi [31] from rising fire plumes, shown in Figure 10.21, find a value of approximately 2. These results apply to plume rise in a tall open space of air at a uniform temperature. The results can be important for issues of fire detection and sprinkler response. Plume rise in a thermally stratified stable ðdT1 =dz > 0Þ atmosphere will not continue indefinitely. Instead, it will slow and eventually stop and form a horizontal layer. It will stop where its momentum becomes zero, roughly when the plume temperature is equal to the local ambient temperature.

10.5.2

Fireball or thermal

The cloud of rising hot gas due to the sudden release of a finite amount of energy is called a thermal. If this occurs due to the sudden release of a finite volume of gaseous fuel that is ignited, the thermal then initially rises as a combusting region. This is depicted by the

TRANSIENT ASPECTS OF FIRE PLUMES

Figure 10.21

329

Rise time of starting fire plumes (from Tanaka, Fujita and Yamaguchi [31])

photographs in Figure 10.22 presented by Fay and Lewis [32]. Before burnout, the combusting region approximates a spherical shape, or fireball. We shall follow the analysis by Fay and Lewis [32] to describe the final diameter of the fireball ðDb Þ, its ultimate height at burnout ðzb Þ and the time for burntout ðtb Þ. Consider the rising fireball

Figure 10.22 Sketches of photographs of a rising ‘fireball’ (from Fay and Lewis [32])

330

FIRE PLUMES

Figure 10.23

Dynamics of a spherical fireball

in a control volume moving with the cloud, as shown in Figure 10.23. The release fuel had a volume VF at temperature T1 and atmospheric pressure p1 . The following approximations are made: 1. The fireball has uniform properties. 2. The relative, normal entrainment velocity is uniform over the fireball and proportional to the rising velocity, dz=dt, i.e. ve ¼ o dz=dt, with an entrainment constant. 3. The pressure on the fireball is constant except for pressure effects due to buoyancy. 4. The actual air entrained is n times the stoichiometric air needed for combustion. By conservation of mass from Equation (3.15), d 4 3 dz Rb ¼ 1 o 4 R2b dt 3 dt

ð10:68Þ

Conservation of energy gives, from Equation (3.48), which is similar to the development of Equation (10.35), 4 dT hc cp R3b þ m_ e cp ðT T1 Þ ¼ ð1 Xr Þm_ e YO2 ;1 3 dt nr

ð10:69Þ

Since the mass of the fireball is small and changes in temperature are likely to be slow, we will ignore the transient term in Equation (10.69). These lead to a constant fireball temperature identical to the fire temperature of Equation (10.35): T T1 ¼

ð1 Xr ÞYO2 ;1 hc nr cp

ð10:70Þ

By the ideal gas law, assuming no change in molecular weight, p ¼ RT=M, or density must also be constant. Consequently, Equation (10.68) becomes

dRb dz ¼ 1 o dt dt

ð10:71Þ

TRANSIENT ASPECTS OF FIRE PLUMES

or integrating with Rb ¼

3

1=3

0 at t ¼ 0: 1=3 T1 3 Vb z Rb ¼ o 1 4 o T 4 Vb

331

ð10:72Þ

We will now apply conservation of vertical momentum. Following Example 3.3, the buoyant force counteracting the weight of the fireball is equal to the weight of ambient atmosphere displaced. For the fireball of volume, Vb ¼ 43 R3b , d dz Vb ¼ ð1 ÞgVb dt dt

ð10:73Þ

where drag effects have been ignored. By substituting for z from Equation (10.72) and integrating, it can be shown that

T T o 1 t2 ð10:74Þ Rb ¼ T1 14 1 The maximum height and diameter for the fire occur when all of the fuel burns out at t ¼ tb . The mass of the fireball at that time includes all of the fuel ðmF Þ and all of the air entrained ðme Þ up to that height. The mass of air entrained can be related to the excess air factor, n, the stoichiometric oxygen to fuel mass ratio, r, and the ambient oxygen mass fraction, YO2 ;1 . Thus the fireball mass at burnout is mb ¼ mF þ me nr ¼ mF 1 þ YO2 ;1

ð10:75Þ

In terms of the original fuel volume, VF , at T1 and p1 , by the perfect gas relationship at constant pressure and approximately constant molecular weights, mb T VF mF T1 nr T ¼ 1þ VF YO2 ;1 T1

Vb ¼

ð10:76Þ

For a given fuel, Equations (10.70), (10.72), (10.74) and (10.76) suggest that the maximum fireball height and diameter behaves as zb

and

1=3

Db V F

ð10:77Þ

and the burnout time follows: 1=6

t b VF

This is supported by the data of Fay and Lewis in Figures 10.24(a) to (c) [32].

ð10:78Þ

332

FIRE PLUMES

Figure 10.24

Fireball dynamics for propane (from Fay and Lewis [32]): (a) flame height, zb , (b) burn-out time, tb , (c) maximum fireball diameter, Db

References 1. Rouse, H., Yih, C.S. and Humphries, H.W., Gravitational convection from a boundary source, Tellus, 1952, 4, 202–10. 2. Morton, B.R., Taylor, G.I. and Turner, J.S. Turbulent gravitational convection from maintained and instanteous sources, Proc. Royal Soc. London, Ser. A, 1956, 234, 1–23.

REFERENCES

333

3. Yokoi, S., Study on the prevention of fire spread caused by hot upward current, Building Research Report 34, Japanese Ministry of Construction, 1960. 4. Steward, F.R., Linear flame heights for various fuels, Combustion and Flame, 1964, 8, 171–78. 5. Steward, F.R., Prediction of the height of turbulent buoyant diffusion flames, Combust. Sci. Technol., 1970, 2, 203–12. 6. McCaffrey, B.J., Purely Buoyant Diffusion Flames: Some Experimental Results, NBSIR 79–1910, National Bureau of Standards, Gaithersburg, Maryland, 1979. 7. Pagni, P.J., Pool fire vortex shedding frequencies, Applied Mechanics Review, 1990, 43, 153–70. 8. Zukoski, E.E., Properties of fire plumes, in Combustion Fundamentals of Fire (ed. G. Cox), Academic Press, London, 1995. 9. Heskestad, G., Fire plumes, in The SFPE Handbook of Fire Protection Engineering, 2nd edn, (eds P.J. DiNenno et al.) Section 2, National Fire Protection Association, Quincy, Massachusetts, 1995, pp. 2-9 to 2-19. 10. Delichatsios, M.A., Air entrainment into buoyant jet flames and pool fires, in The SFPE Handbook of Fire Protection Engineering, 2nd edn (eds P.J. DiNenno et al.), Section 2, National Fire Protection Association, Quincy, Massachusetts, 1995, pp. 2-20 to 2-31. 11. McCaffrey, B., Flame heights, in The SFPE Handbook of Fire Protection Engineering, 2nd edn (eds P.J. DiNenno et al.), Section 2, National Fire Protection Association, Quincy, Massachusetts, 1995, pp. 2-1 to 2-8. 12. Quintiere, J.G. and Grove, B.S., A united analysis for fire plumes, Proc. Comb. Inst., 1998, 27, pp. 2757–86. 13. Yuan, L. and Cox, G., An experimental study of some fire lines, Fire Safety J., 1996, 27, 123–39. 14. Baum, H.R. and McCaffrey, B.J., Fire induced flow field: theory and experiment, in 2nd Symposium (International) on Fire Safety Science, Hemisphere, New York, 1988, pp. 129–48. 15. Koseki, H., Combustion characteristics of hazardous materials, in The Symposium for the NRIFD 56th Anniversary, Mitaka, Tokyo, National Research Institute of Fire and Disaster, 1 June 1998. 16. Smith, D.A. and Cox, G., Major chemical species in buoyant turbulent diffusion flames, Combustion and Flame, 1992, 91, 226–38. 17. Kung, H.C. and Stavrianidis, P., Buoyant plumes of large scale pool fires, Proc. Comb. Inst., 1983, 19, p. 905 18. Cox, G. and Chitty, R., Some source-dependent effects of unbounded fires, Combustion and Flame, 1985, 60, 219–32. 19. Heskestad, G., Virtual origins of fire plumes, Fire Safety J., 1983, 5, 109–14. 20. Hasemi, Y. and Tokunaga, T., Flame geometry effects on the buoyant plumes from turbulent diffusion flames, Combust. Sci. Technol., 1984, 40, 1–17. 21. Ricou, F.P. and Spalding, B.P., Measurements of entrainment by axial symmetric turbulent jets, J. Fluid Mechanics, 1961, 11, 21–32. 22. Heskestad, G., Fire plume air entrainment according to two competing assumptions, Proc. Comb. Inst., 1986, 21, pp. 111–20. 23. Thomas, P.H., Webster, C.T. and Raftery, M.M., Experiments on buoyant diffusion flames, Combustion and Flame, 1961, 5, 359–67. 24. Heskestad, G., A reduced-scale mass fire experiment, Combustion and Flame, 1991, 83, 293– 301. 25. Zukoski, E.E., Fluid dynamics of room fires, in Fire Safety Science, Proceedings of the 1st International Symposium, Hemisphere, Washington, DC, 1986. 26. Hawthorne, W.R., Weddell, D.S. and Hottel, H.C., Mixing and combustion in turbulent gas jets, Proc. Comb. Inst., 1949, 3, pp. 266–88.

334

FIRE PLUMES

27. You, H.Z. and Faeth, G.M., Ceiling heat transfer during fire plume and fire impingment, Fire and Materials, 1979, 3, 140–7. 28. Gross, D., Measurement of flame length under ceilings, Fire Safety J., 1989, 15, 31–44. 29. Ahmad, T. and Faeth, G.M., Turbulent wall fires, Proc. Comb. Inst., 1978, 17, p. 1149–1162. 30. Turner, J. S., The ‘starting plume’ in neutral surroundings, J. Fluid Mechanics, 1962, 13, 356. 31. Tanaka, T., Fujita, T. and Yamaguchi, J., Investigation into rise time of buoyant fire plume fronts, Int. J. Engng Performanced-Based Fire Codes, 2000, 2, 14–25. 32. Fay, J.A. and Lewis, D.H., Unsteady burning of unconfined fuel vapor clouds, Proc. Comb. Inst., 1976, 16, pp. 1397–1405.

Problems 10.1

A 1.5 diameter spill of polystyrene beads burns at 38 g/m2 s. Its flame radiative fraction is 0.54. (a) Compute the centerline temperature at 0.5 m above the base. (b) Compute the centerline temperature at 4.0 m above the base. (c) Compute the flame height.

10.2

Turbulent fire plume temperatures do not generally exceed 1000 C. This is well below the adiabatic flame temperature for fuels. Why?

10.3

A heptane pool fire of diameter 10 m burns at 66 g/m2 s. Its radiative fraction is 0.10 and the heat of combustion is 41 kJ/g. (a) Calculate its flame height. (b) Calculate its centerline temperature at a height of twice its flame length.

10.4

Given: an axisymmetric fire burns at the following rates:

Diameter (D) in m

m_ 00 in g/m2 s

0.5 1.0 2.0

20 45 60 R

= 25

H = 3m

D

PROBLEMS

335

The fuel heat of combustion is 40 kJ/g and its radiative fraction (Xr) is 0.30. For each diameter, compute the following: (a) Gas temperature at the ceiling above the fire (maximum). (b) Maximum velocity impinging on the ceiling above the fire. (c) The extent of the flame, its height and any interaction with the ceiling. (d) The maximum radiant heat flux to a target a radius r from the center of the pool fire can be estimated by Xr Q_ =ð4 r 2 Þ. Compute the heat flux at 3 m. 10.5

For a constant energy release rate fire, will its flame height be higher in a corner or in an open space? Explain the reason for your answer.

10.6

Assume that the diameter of the fire as small and estimate the temperature at the ceiling when the flame height is (a) 1/3, (b) 2/3 and (c) equal to the height of the ceiling. The fire is at the floor level and the distance to the ceiling is 3 m. Also calculate the corresponding energy release rates of the fire for (a), (b) and (c). The ambient temperature, T1 ¼ 25 C.

T∞ = 25 °C H = 3m zf D

10.7

A steel sphere is placed directly above a fire source of 500 kW. The height of the sphere is 3 m. Its properties are given below:

3m . Q =500 kW

Initial temperature, To ¼ 20 C. Specific heat, cp ¼ 460 J=kg K Density, ¼ 7850 kg=m3 Radius, R ¼ 1 cm Volume of sphere ¼ (4/3) R3 Suface area of sphere ¼ 4 R2

336

FIRE PLUMES (a) What is the gas temperature at the sphere? (b) What is the gas velocity at the sphere? (c) How long will it take for the sphere to reach 100 C? Assume convective heating only with a constant heat transfer coefficient, h ¼ 25 W=m2 K.

10.8

A 1 MW fire occurs on the floor of a 15 m high atrium. The diameter of the fire is 2 m. A fast response sprinkler link at the ceiling directly above the fire will activate when the gas temperature reaches 60 C. Will it activate? Explain your answer. The ambient temperature is 20 C.

15 m

10.9

A sprinkler is 8 m directly above a fire in an atrium. It requires a minimum temperature rise of 60 C to activate. The fire diameter is 2 m and the flame radiative fraction is 40 %.

Sprinkler 8m

D=2

(a) Assuming no heat loss from the sprinkler, calculate the minimum energy release rate needed for sprinkler activation. (b) What is the flame height? (c) Estimate the maximum radiation heat flux incident to an object 6 m away. 10.10

A grease pan fire occurs on a stove top. The flames just touch the ceiling. The pan is 1 m above the floor and the ceiling is 2.4 m high. The pan is 0.3 m in diameter.

PROBLEMS

337

4m

T∞ = 20 °C

2.4 m

1m

Stove

(a) Estimate the energy release rate of the fire. (b) A sprinkler designed to activate immediately when the local temperature reaches 80 C is located 4 m away. Will it activate? Show results to explain. 10.11 It is estimated that in the WTC disaster that the jet fuel burned in one of the towers at 10 GW. If this is assumed to burn over an equivalent circular area constituting one floor of 208 208 ft, and would produce an unimpeded fire plume, then compute the height of the flame and the temperature at two diameters above the base. 10.12 The aircraft that hit one of the WTC towers was carrying about 28 500 kg of JP4 liquid aviation fuel. Three fireballs were seen following impact. They achieved a maximum diameter of 60 m each. NIST estimates that 10–25 % of the fuel burned in the fireballs. Make a quantitative estimate of the fuel that burned in these fireballs and see if NIST is correct. Take the density of the JP4 vapor at 20 C to be 2.5 kg/m3. Its liquid density is 760 kg/m3.

11 Compartment Fires

11.1 Introduction The subject of compartment fires embraces the full essence of fire growth. The ‘compartment’ here can represent any confined space that controls the ultimate air supply and thermal environment of the fire. These factors control the spread and growth of the fire, its maximum burning rate and its duration. Although simple room configurations will be the limit of this chapter, extensions can be made to other applications, such as tunnel, mine and underground fires. Spread between compartments will not be addressed, but are within the scope of the state-of-the-art. A more extensive treatment of enclosure fires can be found in the book by Karlsson and Quintiere [1]. Here the emphasis will be on concepts, the interaction of phenomena through the conservation equations, and useful empirical formulas and correlations. Both thermal and oxygenlimiting feedback processes can affect fire in a compartment. In the course of fire safety design or fire investigation in buildings, all of these effects, along with fire growth characteristics of the fuel, must be understood. The ability to express the relevant physics in approximate mathematics is essential to be able to focus on the key elements in a particular situation. In fire investigation analyses, this process is useful to develop, first, a qualitative description and then an estimated time-line of events. In design analyses, the behavior of fire growth determines life safety issues, and the temperature and duration of the fire controls the structural fire protection needs. While such design analysis is not addressed in current prescriptive regulations, performance codes are evolving in this direction. The SFPE Guide to Fire Exposures is an example [2]. Our knowledge is still weak, and there are many features of building fire that need to be studied. This chapter will present the basic physics. The literature will be cited for illustration and information, but by no means will it constitute a complete review. However, the work of Thomas offers much, e.g. theories (a) on wood crib fires [3], (b) on flashover and instabilities [4,5], and (c) on pool fires in compartments [6].

Fundamentals of Fire Phenomena James G. Quintiere # 2006 John Wiley & Sons, Ltd ISBN: 0-470-09113-4

340

COMPARTMENT FIRES

Figure 11.1

11.1.1

Compartment fire scenarios

Scope

The scope of this chapter will focus on typical building compartment fires representative of living or working spaces in which the room height is nominally about 3 m. This scenario, along with others, is depicted in Figure 11.1, and is labeled (a). The other configurations shown there can also be important, but will not necessarily be addressed here. They include: (b) tall, large spaces where the contained oxygen reservoir is important; (c) leaky compartments where extinction and oscillatory combustion have been observed (Tewarson [7], Kim, Ohtami and Uehara [8], Takeda and Akita [9]); (d) compartments in which the only vent is at the ceiling, e.g. ships, basements, etc. Scenario (a) will be examined over the full range of fire behavior beginning with spread and growth through ‘flashover’ to its ‘fully developed’ state. Here, flashover is defined as a transition, usually rapid, in which the fire distinctly grows bigger in the compartment. The fully developed state is where all of the fuel available is involved to its maximum extent according to oxygen or fuel limitations.

11.1.2

Phases of fires in enclosures

Fire in enclosures may be characterized in three phases. The first phase is fire development as a fire grows in size from a small incipient fire. If no action is taken to suppress the fire, it will eventually grow to a maximum size that is controlled by the

INTRODUCTION

Figure 11.2

341

Phases of fire development

amount of fuel present (fuel controlled) or the amount of air available through ventilation openings (ventilation limited). As all of the fuel is consumed, the fire will decrease in size (decay). These stages of fire development can be seen in Figure 11.2. The fully developed fire is affected by (a) the size and shape of the enclosure, (b) the amount, distribution and type of fuel in the enclosure, (c) the amount, distribution and form of ventilation of the enclosure and (d) the form and type of construction materials comprising the roof (or ceiling), walls and floor of the enclosure. The significance of each phase of an enclosure fire depends on the fire safety system component under consideration. For components such as detectors or sprinklers, the fire development phase will have a great influence on the time at which they activate. The fully developed fire and its decay phase are significant for the integrity of the structural elements. Flashover is a term demanding more attention. It is a phenomenon that is usually obvious to the observer of fire growth. However, it has a beginning and an end; the former is the connotation for flashover onset time given herein. In general, flashover is the transition between the developing fire that is still relatively benign and the fully developed fire. It usually also marks the difference between the fuel-controlled or well-ventilated fire and the ventilation-limited fire. The equivalence ratio is less than 1 for the former and greater than 1 for the latter, as it is fuel-rich. Flashover can be initiated by several mechanisms, while this fire eruption to the casual observer would appear to be the same. The observer would see that the fire would ‘suddenly’ change in its growth and progress to involving all of the fuel in the compartment. If the compartment does not get sufficient stoichiometric air, the fire can produce large flames outside the compartment. A ventilation-limited fire can have burning mostly at the vents, and significant toxicity issues arise due to the incomplete combustion process. Mechanisms of flashover can include the following: 1. Remote ignition. This is the sudden ignition by autoignition or piloted ignition, due to flaming brands, as a result of radiant heating. The radiant heating is principally from the compartment ceiling and hot upper gases due to their large extent. The threshold for the piloted ignition of many common materials is about 20 kW/m2. This value of flux, measured at the floor, is commonly taken as an operational criterion for flashover. It also corresponds to gas temperatures of 500–600 C.

342

COMPARTMENT FIRES

2. Rapid flame spread. As we know, radiant preheating of a material can cause its surface temperature to approach its piloted ignition temperature. This causes a singularity in simple flame spread theory that physically means that a premixed mixture at its lower flammability limit occurs ahead of the surface flame. Hence, a rapid spread results in the order of 1 m/s. 3. Burning instability. Even without spread away from a burning item, a sudden fire eruption can be recognized under the right conditions. Here, the thermal feedback between the burning object and the heated compartment can cause a ‘jump’ from an initial stable state of burning at the unheated, ambient conditions, to a new stable state after the heating of the compartment and the burning of the fuel come to equilibrium. 4. Oxygen supply. This mechanism might promote back-draft. It involves a fire burning in a very ventilation-limited state. The sudden breakage of a window or the opening of a door will allow fresh oxygen to enter along the floor. As this mixes with the fuel-rich hot gases, a sudden increase in combustion occurs. This can occur so rapidly that a significant pressure increase will occur that can cause walls and other windows to fail. It is not unlike a premixed gas ‘explosion’. 5. Boilover. This is a phenomenon that occurs when water is sprayed on to a less dense burning liquid with a boiling temperature higher than that of water. Water droplets plunging through the surface can become ‘instant’vapor, with a terrific expansion that causes spraying out of the liquid fuel. The added increase of area of the liquid fuel spray can create a tremendous increase in the fire. Even if this does not happen at the surface, the collection of the heavier water at the bottom of a fuel tank can suddenly boil as the liquid fuel recedes to the bottom. Hence, the term ‘boilover’.

11.2 Fluid Dynamics The study of fire in a compartment primarily involves three elements: (a) fluid dynamics, (b) heat transfer and (c) combustion. All can theoretically be resolved in finite difference solutions of the fundamental conservation equations, but issues of turbulence, reaction chemistry and sufficient grid elements preclude perfect solutions. However, flow features of compartment fires allow for approximate portrayals of these three elements through global approaches for prediction. The ability to visualize the dynamics of compartment fires in global terms of discrete, but coupled, phenomena follow from the flow features.

11.2.1

General flow pattern

The stratified nature of the flow field due to thermally induced buoyancy is responsible for most of the compartment flow field. Figure 11.3 is a sketch of a typical flow pattern in a compartment. Strong buoyancy dominates the flow of the fire. Turbulence and pressure cause the ambient to mix (entrainment) into the fire plume. Momentum and thermal buoyancy

FLUID DYNAMICS

Figure 11.3

343

Flow pattern in a compartment fire

causes a relatively thin ceiling jet to scrub the ceiling. Its thickness is roughly one-tenth of the room height. A counterpart cold floor jet also occurs. In between these jets, recirculating flows form in the bulk of the volume, giving rise to a four-layer flow pattern. At the mid-point of this four-layer system is a fairly sharp boundary (layer interface) due to thermal stratification between the relatively hot upper layer and cooler lower layer. Many of these flow attributes are displayed in photographs taken by Rinkinen [10] for ‘corridor’ flows, made visible by white smoke streaks Figure 11.4. Figure 11.5 shows the sharp thermal stratification of the two layers in terms of the gas or wall temperatures [11].

11.2.2

Vent flows

The flow of fluids through openings in partitions due to density differences can be described in many cases as an orifice-like flow. Orifice flows can be modeled as inviscid Bernoulli flow with an experimental correction in terms of a flow coefficient, C, which generally depends on the contraction ratio and the Reynolds number. Emmons [12] describes the general theory of such flows. For flows through horizontal partitions, at near zero pressure changes, the flow is unstable and oscillating bidirectional flow will result. In this configuration, there is a flooding pressure above which unidirectional Bernoulli flow will occur. Epstein [13] has developed correlations for flows through horizontal partitions. Steckler, Quintiere and Rinkinen [14] were the first to measure successfully the velocity and flow field through vertical partitions for fire conditions. An example of their measured flow field is shown in Figure 11.6 for a typical module room fire. The pressure difference promoting vertical vent flows are solely due to temperature stratification, and conform to a hydrostatic approximation. In other words, the momentum equation in the vertical direction is essentially, due to the low flows: @p ¼ g @z

ð11:1Þ

344

COMPARTMENT FIRES

Figure 11.4

(a) Smoke layers in a model corridor as a function of the exit door width (b) Smoke streaklines showing four directional flows (c) Smoke streaklines showing mixing of the inlet flow from the right at the vent with the corridor upper layer [10]

FLUID DYNAMICS

345

Figure 11.5 Thermal layers [11]

From Figure 11.3, under free convection, there will be a height in the vent at which the flow is zero, N; this is called the neutral plane. The pressure difference across the vent from inside (i) to outside (o) can be expressed above the neutral plane as pi po ¼

Z

z

ðo i Þg dz

0

Figure 11.6

Measured flow field in the doorway of a room fire [14]

ð11:2Þ

346

COMPARTMENT FIRES

Figure 11.7

Zone modeling concepts

Such pressure differences are of the order of 10 Pa for z 1 m and therefore these flows are essentially at constant pressure (1 atm 105 Pa). The approach to computing fire flows through vertical partitions is to: (a) compute velocity from Bernoulli’s equation using Equation (11.2), (b) substitute temperatures for density using the perfect gas law and, finally, (c) integrate to compute the mass flow rate. This ideal flow rate is then adjusted by a flow coefficient generally found to be about 0.7 [12,14]. Several example flow solutions for special cases will be listed here. Their derivations can be found in the paper by Rockett [15]. The two special cases consider a two-zone model in which there is a hot upper and lower cool homogeneous layer as shown in Figure 11.7(a); this characterizes a developing fire. Figure 11.7(b) characterizes a large fully developed fire as a single homogeneous zone. The two-zone model gives the result in terms of the neutral plane height ðHn Þ and the layer height ðHs Þ for a doorway vent of area Ao and height Ho . The ambient to room temperature ratio is designated as To =T: h i3=2 pﬃﬃﬃﬃﬃﬃ pﬃﬃﬃﬃﬃ n m_ out ¼ 23 2gCo Ao Ho ½ðÞð1 ðÞÞ1=2 1H Ho

ð11:3Þ

1=2 pﬃﬃﬃﬃﬃﬃ pﬃﬃﬃﬃﬃ Hs Hn Hs n m_ in ¼ 23 2gCo Ao Ho ½1 ðÞ1=2 H þ Ho Ho Ho 2Ho

ð11:4Þ

The inflow is generally the ambient air and the outflow is composed of combustion products and excess air or fuel.

HEAT TRANSFER

347

For a one-zone model under steady fire conditions, the rate mass of air inflow can be further computed in terms of the ratio of fuel to airflow rate supplied ðÞ as m_ air

pﬃﬃﬃﬃﬃﬃpﬃﬃﬃpﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ Ho g 2ð1 Þ ¼ h i3=2 1 þ ½ð1 þ Þ2 =1=3 2 3 o CAo

ð11:5Þ

It is interesting to contrast Equation (11.5) with the air flow rate at a horizontal vent under bidirectional, unstable flow from Epstein [13]:

m_ air ¼

5=4 pﬃﬃﬃpﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

0:0681 Ao

g

ð1 þ Þ

2ð1 Þ

1=2

ð11:6Þ

This horizontal vent flow is nearly one-tenth of that for the same size of vertical vent. Thus, a horizontal vent with such unstable flow, as in a basement or ship fire, is relatively inefficient in supplying air to the fire. A useful limiting approximation for the air flow rate into a vertical vent gives the maximum flow as pﬃﬃﬃﬃﬃﬃ m_ air ¼ ko Ao Ho ;

where ko ¼ 0:5 kg=s m5=2

ð11:7Þ

This holds for < 23, and Hs small. In general, it is expected that Hn =Ho ranges from 0.3 to 0.7 and Hs =Ho from 0 to 0.6 for doorways [15].

11.3 Heat Transfer The heat transfer into the boundary surface of a compartment occurs by convection and radiation from the enclosure, and then conduction through the walls. For illustration, a solid boundary element will be represented as a uniform material having thickness, , thermal conductivity, k, specific heat, c, and density, . Its back surface will be considered at a fixed temperature, To . The heat transfer path through the surface area, A, can be represented as an equivalent electric circuit as shown in Figure 11.8. The thermal resistances, or their inverses, the conductances, can be computed using standard heat transfer methods. Some will be illustrated here.

Figure 11.8

Wall heat transfer

348

COMPARTMENT FIRES

11.3.1

Convection

Convections to objects in a fire environment usually occur under natural convection conditions. Turbulent natural convection is independent of scale and might be estimated from hc H ¼ 0:13 Nu ¼ K

1=3 gðT To ÞH 3 Pr T 2

ð11:8aÞ

where is the kinematic viscosity and Pr is the Prandt number. It gives hc of about 10 W/m2 K. Under higher flow conditions, it is possible that hc might be as high as 40 W/m2 K. This has been shown in measurements by Tanaka and Yamada [16] as the floor-to-ceiling coefficients vary from about 5 to 40 W/m2 K. They found that an empirical correlation for the average compartment convection coefficient is hc 2:0 103 ; QH 4 103 pﬃﬃﬃﬃﬃﬃﬃ ¼ 2=3 1 cp gH 0:08QH ; QH > 4 103

ð11:8bÞ

QH is based on height, H (see Equation (10.41)).

11.3.2

Conduction

Only a finite difference numerical solution can give exact results for conduction. However, often the following approximation can serve as a suitable estimation. For the unsteady case, assuming a semi-infinite solid under a constant heat flux, the exact solution for the rate of heat conduction is rﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ kc q_ w ¼ A ðTw To Þ 4 t

ð11:9Þ

rﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ kc hk ¼ 4 t

ð11:10Þ

or, following Figure 11.8,

For steady conduction, the exact result is hk ¼

k

ð11:11Þ

The steady state result would be considered to hold for time, t> approximately, from Equation (7.35).

2 4½k=c

ð11:12Þ

HEAT TRANSFER Table 11.1

349

Approximate thermal properties for typical building material

k (W/m K) kc (W2 s/m4 K2) k=c (m2/s)

Concrete/brick

Gypsum

1 106 5 107

0.5 105 4 107

Mineral wool 0.05 103 5 107

Table 11.1 gives some typical properties. An illustration for a wall six inches thick, or 0:15 m, gives, for most common building materials, t¼

2 ¼ 3:1 hours 4½k=c

from Equation (11.12) for the thermal penetration time. Hence, most boundaries might be approximated as thermally thick since compartment fires would typically have a duration of less than 3 hours. Since the thermally thick case will predominate under most fire and construction conditions, the conductance can be estimated from Equation (11.10). Values for the materials characteristic of Table 11.1 are given in Table 11.2. As time progresses, the conduction heat loss decreases.

11.3.3

Radiation

Radiation heat transfer is very complex and depends on the temperature and soot distribution for fundamental computations. Usually, these phenomena are beyond the state-of-the-art for accurate fire computations. However, rough approximations can be made for homogeneous gray gas approximations for the flame and smoke regions. Following the methods in common texts (e.g. Karlsson and Quintiere [1]) formulas can be derived. For example, consider a small object receiving radiation in an enclosure with a homogeneous gray gas with gray uniform walls, as portrayed in Figure 11.9. It can be shown that the net radiation heat transfer flux received is given as h i 00 q_ r ¼ ðTg4 T 4 Þ wg ðTg4 Tw4 Þ and

wg ¼

Table 11.2 tðminÞ 10 30 120

ð1 g Þ w

w þ ð1 w Þ g

Typical wall conductance values hk ðW=m2 KÞ 0.8–26 0.3–10 0.2–5

ð11:13Þ

350

COMPARTMENT FIRES

Figure 11.9

Radiation of a small object in an enclosure

where

¼ emissivity of the target

w ¼ emissivity of the wall

g ¼ emissivity of the gas T ¼ target temperature Tw ¼ wall temperature Tg ¼ gas temperature and ¼ Stefan–Boltzmann constant ¼ 5:67 1011 kW/m2K4 If the object is the wall itself, then Equation (11.13) simplifies to the rate of radiation received by the wall as q_ r ¼

AðT 4 Tw4 Þ 1= g þ 1= w 1

ð11:14Þ

Since the boundary surface will become soot-covered as the fire moves to a fully developed fire, it might be appropriate to set w ¼ 1. The gas emissivity can be approximated as

g ¼ 1 eH

ð11:15Þ

where H represents the mean beam length for the enclosure that could be approximated as its height. The absorption coefficient of the smoke or flames, , could range from about 0.1 to 1 m1. For the smoke conditions in fully developed fires, ¼ 1 m1 is a reasonable estimate, and hence g could range from about 0.5 for a small-scale laboratory enclosure to nearly 1 for building fires.

351

HEAT TRANSFER

Under fully developed conditions, the radiative conductance can be expressed as hr ¼ g ðT 2 þ Tw2 ÞðT þ Tw Þ

ð11:16Þ

It can be estimated, for g ¼ 1 and T ¼ Tw , that hr ¼ 104 – 725 W/m2 K for T ¼ 500 – 1200 C.

11.3.4

Overall wall heat transfer

From the circuit in Figure 11.7, the equivalent conductance, h, allows the total heat flow rate to be represented as q_ w ¼ hAðT To Þ

ð11:17aÞ

1 1 1 ¼ þ h hc þ hr hk

ð11:17bÞ

where

For a fully developed fire, conduction commonly overshadows convection and radiation; therefore, a limiting approximation is that h hk , which implies Tw T. This result applies to structural and boundary elements that are insulated, or even to concrete structural elements. This boundary condition is ‘conservative’ in that it gives the maximum possible compartment temperature.

11.3.5

Radiation loss from the vent

From Karlsson and Quintiere [1], it can be shown that for an enclosure with blackbody surfaces ð w ¼ 1Þ, the radiation heat transfer rate out of the vent of area Ao is q_ r ¼ Ao g ðTg4 To4 Þ þ Ao ð1 g ÞðTw4 To4 Þ

ð11:18Þ

Since, for large fully developed fires, g is near 1 or Tw Tg, then it follows that q_ r ¼ Ao ðT 4 To4 Þ

ð11:19Þ

and shows that the opening acts like a blackbody. This blackbody behavior for the vents has been verified, and is shown in Figure 11.10 for a large data set of crib fires in enclosures of small scale with H up to 1.5 m [17]. The total heat losses in a fully developed fire can then be approximated as q_ ¼ q_ w þ q_ r ¼ hk AðT To Þ þ Ao ðT 2 þ To2 ÞðT þ To ÞðT To Þ

ð11:20Þ

352

COMPARTMENT FIRES

Figure 11.10 Radiation from the compartment vent [17]

11.4 Fuel Behavior A material burning in an enclosure will depart from its burning rate in normal air due to thermal effects of the enclosure and the oxygen concentration that controls flame heating. Chapter 9 illustrated these effects in which Equation (9.73) describes ‘steady’ burning in the form: 00

q_ m_ ¼ L 00

ð11:21Þ

If the fuel responds fast to the compartment changes, such a ‘quasi-steady’ burning rate model will suffice to explain the expenditure of fuel mass in the compartment. The fuel heat flux is composed of flame and external (compartment) heating. The flame temperature depends on the oxygen mass fraction ðYO2 Þ, and external radiant heating depends on compartment temperatures.

11.4.1

Thermal effects

The compartment net heat flux received by the fuel within the hot upper layer for the blackbody wall and fuel surfaces can be expressed from Equation (11.13) as 00

q_ r ¼ g g ðTg4 Tv4 Þ þ ð1 g ÞðTw4 Tv4 Þ

ð11:22Þ

where Tv is the vaporization temperature of the fuel surface. For fuel in the lower layer, an appropriate view factor should reduce the flux given in Equation (11.22).

Mass Loss Rate per unit Total Area (g/m2-s)

FUEL BEHAVIOR

Figure 11.11

353

80 70 60 50 40 30 20

PMMA Pools Vent limited

10

PU Cribs Wood Cribs 0 200 400 600 800 1000 1200 Compartment Ceiling Gas Temperature (°C)

0

Compartment thermal effect on burning rate for wood cribs and pool fire

The quasi-steady maximum burning rate is illustrated in Figure 11.11 for wood cribs and PMMA pool fires under various fuel loading and ventilation conditions [18,19]. The cribs represent fuels controlled by internal combustion effects and radiant heat transfer among the sticks. The pool fires shown in Figure 11.11 are of small scale and have low absorptivity flames. Consequently, the small pools represent a class of fuels that are very responsive to the thermal feedback of the compartment. In general, the small-scale ‘pool’ fires serve to represent other fuel configurations that are equally as responsive, such as surfaces with boundary layer or wind-blown flames. The crib and pool fire configurations represent aspects of realistic building fuels. They offer two extremes of fuel sensitivity to compartment temperature.

11.4.2

Ventilation effects

As the vent is reduced, mixing will increase between the two layers and the oxygen feeding the fire will be reduced, and the burning rate will correspondingly be reduced. This is shown for some of the data in Figure 11.11 and represents ventilation-controlled burning. Figure 11.12 shows more dramatic effects of ventilation for experimental heptane pool fires in a small-scale enclosure [20]. These experiments included the case of a single ceiling vent or two equal-area upper and lower wall vents. For vent areas below the indicated flame extinction boundary, the fire self-extinguishes. For a fuel diameter of 95 mm, the mass loss rate associated with the ceiling vent was much lower than the wall vent case. The air supply differences indicated by Equations (11.5) and (11.6) suggest that the fuel mass loss rate for the ceiling vent case was limited by air flow.

354

COMPARTMENT FIRES Wall Vent, Fuel 65 mm Wall Vent, Fuel 95 mm Ceiling Vent, Fuel 65 mm Ceiling Vent, Fuel 95 mm Ceiling Vent, Fuel 120 mm

Mass Loss Rate per Area (g/m2s)

80 70

Extinction Boundary 60 per Pan Diameter

120

95

50

65

40 30 20

Free burn 120 mm 95 mm 65 mm

10 0

1

10 100 Total Vent Area (cm2)

1000

Figure 11.12 Ventilation-controlled burning for ceiling and wall vents [20]

11.4.3

Energy release rate (firepower)

The energy release rate of the fire in a compartment may occur within the compartment, and outside due to normal flame extension or insufficient air supply to the compartment. The energy release rate has been commonly labeled as the heat release rate (HRR); herein the term ‘firepower’ is fostered, as that is the common term adopted in energy applications. Moreover, the energy associated with chemical reaction is not heat in strict thermodynamic terminology. However, it is recognized that the fire community usage is the ‘heat release rate’. Nevertheless, the important factor in compartment fire analysis is to understand the conditions that allow for burning within and outside a compartment. The heat of the flames and smoke causes the fuel to vaporize at a mass flow rate, m_ F . While all of the fuel may eventually burn, it may not necessarily burn completely in the compartment. This depends on the air supply rate. Either all of the fuel is burned or all of the oxygen in the incoming air is burned. That which burns inside gives the firepower within the enclosure. Thus, the firepower within the enclosure is given as Q_ ¼

m_ F hc ; m_ air hair ;

<1 1

ð11:23Þ

where m_ air is the mass flow rate of air supplied to the compartment and hair is the heat of combustion per unit mass of air, an effective constant for most fuels at roughly 3 kJ/g air. The firepower is based on burning all of the gaseous fuel supplied or all of the air. The equivalence ratio, , determines the boundary between the fuel-lean and

ZONE MODELING AND CONSERVATION EQUATIONS

355

fuel-rich combustion regimes: ¼

sm_ F m_ air

ð11:24Þ

where s is the stoichiometric air to fuel ratio. Stoichiometry is generally not obvious for realistic fuels as they do not burn completely. However, s can be computed from s¼

hc hair

ð11:25Þ

where hc is the heat of combustion (the chemical heat of combustion, as given in Table 2.3). It is important to distinguish between the mass loss or supply rate of the fuel and its burning rate within the compartment. The mass loss rate in contrast to Equation (11.23) is given as 8 00 <1 < m_ F;b AF ; m_ F ¼ m_ air F q_ 00r : þ ; 1 s L

ð11:26Þ

00

where m_ F;b is the fuel burn flux, AF is the available fuel area, and F q_ r is the view-factormodified net radiant flux received (Equation (11.22)). The actual area burning in a ventilation controlled fire ð > 1Þ would generally be less than the available area, as suggested by m_ air 00 ¼ m_ F;b AF;b s

ð11:27Þ

A typical transition to behavior of a ventilation-controlled fire begins with excess air as the fire feeds on the initial compartment air, but then is limited by air flow at the vent. As a consequence the fire can move to the vent and withdraw as the fuel is consumed. This might lead to two areas of ‘deep’ burning if the fire is extinguished before complete burnout.

11.5 Zone Modeling and Conservation Equations The above discussion lays out the physics and chemical aspect of the processes in a compartment fire. They are coupled phenomena and do not necessarily lend themselves to exact solutions. They must be linked through an application of the conservation equations as developed in Chapter 3. The ultimate system of equations is commonly referred to as ‘zone modeling’ for fire applications. There are many computer codes available that represent this type of modeling. They can be effective for predictions if the

356

COMPARTMENT FIRES

Figure 11.13

Control volume for conservation equations

fire processes are completely and accurately represented. Complete finite difference modeling of the fundamental equations is a step more refined than zone modeling, and it also has limitations, but will not be discussed here. Zone modeling is no more than the development of the conservation laws for physically justifiable control volumes. The control volume is principally the hot gas layer of the compartment fire. The fire is embedded within this volume. Figure 11.13 can be used to define a control volume for either the two-zone or one-zone approach. In the former, the lower layer would constitute another control volume. For simplicity, the one-zone model will only be considered here for illustration. This limited development is simply to present the methodology and demonstrate the basis for empirical correlations to follow.

11.5.1

Conservation relationships

Principally, conservation of energy for the compartment provides the important relationship to establish the extent of thermal feedback to the fuel. Conservation of mass and oxygen provide additional support equations. The ‘process’ relationships, given previously, establish the important transport rates of mass and energy. These ‘constitutive’ relationships may not always be complete enough to describe all fire scenarios. The conservation equations can be developed from Chapter 3. They are listed below: 1. Mass. For the one-zone model with a perfect gas using Equation (3.15), the conservation of mass can be written as

pV dT þ m_ ¼ m_ air þ m_ F RT 2 dt

ð11:28Þ

The conservation of oxygen for the mass fraction designation by Y follows from Equation (3.22). 2. Oxygen dðmYÞ 0:233sQ_ þ m_ Y m_ air ð0:233Þ ¼ dt hc

ð11:29Þ

357

ZONE MODELING AND CONSERVATION EQUATIONS

Under steady conditions, or where the transient term is negligible, this is simply Y¼

0:233ð1 Þ for 1 1 þ =s

ð11:30Þ

and Y¼0

for the ðglobalÞ equivalence ratio; > 1

The energy equation follows from Equation (3.45) where the ‘loss’ terms are grouped to include both heat and enthalpy transport rates as Q_ l : 3. Energy cv

dðmTÞ cv V dp ¼ ¼ Q_ Q_ 1 dt R dt

ð11:31Þ

The transient term can be neglected in most cases, but any sudden change in the energy will result in a pressure change. This can be responsible for ‘puffing’ effects seen in compartment fires, especially when the burning is oscillatory.

11.5.2

Dimensionless factors in a solution

By normalizing the steady equation with the air flow rate and representing all of the losses as linear in T (Equation (11.31)), the temperature rise can be solved as T To ¼

Q_ =ðcp m_ air Þ xo þ xw þ xr

ð11:32Þ

The dimensionless x-loss factors are parameters that effect temperature, and occur in modified forms in many correlations in the literature for compartment fire temperature. 1. Flow-loss factor. The flow factor, xo , is the ratio of out-flow to airflow: xo ¼ 1 þ

m_ F ¼1þ s m_ air

ð11:33Þ

and is approximately 1 for < 1 since s ranges from 3 to 13 for charring to liquid fuels respectively. 2. Wall-loss factor. The wall-loss factor, xw , is given in terms of the compartment heat transfer surface area, A, and the overall heat transfer coefficient, h, as given by Equation (11.17): xw ¼

hA hA pﬃﬃﬃﬃﬃﬃ m_ air cp ko cp Ao Ho

ð11:34Þ

Here, the maximum air flow rate could be inserted per Equation (11.7). In general, h depends on convection ðhc Þ, radiation ðhr Þ and conduction into the walls ðhk Þ. Their typical values range as hc 10–30, hr 5–100 and hk ðkc=tÞ1=2 5–60 W/m2 K; consequently, h 1 to 100 W/m2 K is a typical range expected. It is important to note

358

COMPARTMENT FIRES

that the wall conductance will decrease with time for thick walls, leading to an increase in temperature. 3. Vent-radiation factor. As shown by Figure 11.10, the radiation flux from the vent is reasonably represented as a blackbody at the gas temperature, T. Thus, the vent radiation loss can be estimated as 00

q_ r;o xr ¼ m_ air cp ðT To Þ

ðT 4 To4 ÞAo T 3 pﬃﬃﬃﬃﬃﬃ pﬃﬃﬃﬃﬃﬃ cp ko Ao Ho ðT To Þ cp ko Ho

for To =T small

ð11:35Þ

This factor might range from roughly 102 to 101 for T ranging from 25 to 1200 C respectively. It is the least significant of the three factors.

11.6 Correlations While computer models exist to solve for many aspects of compartment fires, there is nothing more valuable than the foundation of experimental results. Empirical correlations from such experimental data and based on the appropriate dimensionless groups have been developed by many investigators. These can be very useful to the designer and investigator. As with computer models, these correlations can also be incomplete; they may not have tested over a sufficient range of variables, they may leave out important factors and they are still usually specific to the fuels and their configuration. Nevertheless, they are very powerful, in their algebraic or graphical simplicity, and in their accuracy. Some will be listed here. The SFPE guide addressing fully developed enclosure fires is an excellent source of most correlations [2].

11.6.1

Developing fires

The developing fire applies up to flashover, and therefore might apply to an average temperature in the upper layer of 600 C, at most. McCaffrey, Quintiere and Harkleroad [21] developed a well-known correlation for this domain known as the MQH correlation. The MQH correlation was developed from a large database for compartment fires ranging in scale from 0.3 to 3 m high, and with fuels centered on the floor [21]. Other investigators have followed the form of this correlation for floor fires near a wall, in a corner and for burning wall and ceiling linings [22–25]. This correlation takes the form of Equation (11.32), but as a power law. The dimensionless groups that pertain are dimensionless mass and energy flow rates of the form: m ¼

m_ pﬃﬃﬃ pﬃﬃﬃﬃﬃﬃ o g Ao H o

and

Q ¼

Q_ pﬃﬃﬃ pﬃﬃﬃﬃﬃﬃ o cp To gAo Ho

ð11:36Þ

The dimensionless fuel supply and air inflow rate can be described, accordingly, as mF and mA respectively. The entrainment rate depends on the height over which entrainment

CORRELATIONS

359

Table 11.3 Values of CT for different fire configurations Fire Configuration

CT

Discrete, centered

480 686 615 804 856 1200 1060–1210 1000 940

Discrete, against wall Discrete, in corner Linings, wall only Linings, wall and ceiling

Source McCaffrey et al.[21] Azhakesan et al.[23] Mowrer et al.[22] Azhakesan et al.[23] Mowrer et al.[22] Azhakesan et al. [25] Karlsson [24] Azhakesan et al. [25] Karlsson [24]

occurs Hs , and on the fire configuration. This factor is simply represented here by G. Thus, the entrainment rate and air flow rate might functionally be represented as mA

Hs ¼ function G; Ho

ð11:37Þ

Eliminating the Hs dependence by Equation (11.4), the temperature of the hot layer can generally be functionally expressed T ¼ function Qo ; Qw ; mF ; G To

ð11:38Þ

The MQH correlation for the layer temperature rise has found the empirical fit to data: 2=3

1=3

T ¼ CT ðG; mF ÞQo Qw

ð11:39Þ

where Qo from Equation (11.36) has Q_ as the firepower, and Qw has Q_ taken as hk ATo from Equations (11.10) and (11.11). These Q values correspond to xo and xw in Equation (11.32); i.e. Qo ¼

Q_ pﬃﬃﬃ pﬃﬃﬃﬃﬃﬃ 1 cp T1 gAo Ho

and

Qw ¼

hk A pﬃﬃﬃ pﬃﬃﬃﬃﬃﬃ 1 cp gAo Ho

CT is an empirical constant that appears to increase as the entrainment rate decreases, as shown from the evidence in Table 11.3. Also from Equation (11.32), the temperature will decrease as the dimensionless fuel supply rate increases. Equation (11.39) holds for the overventilation regime, and therefore it should not be used at a point where m_ F =m_ air ¼ 1=s or ¼ 1 without at least modifying Qo . Figure 11.14 shows the original MQH data fit and Figures 11.15 and 11.16 show results for compartment lining material fires. Various lining materials were used and the firepower was measured by oxygen calorimetry. Departure from the linear slope behavior marks the onset of the ventilation-limited fires where the correlation based on the total firepower, within and outside the compartment, does not hold. The results for the lining fires also show the importance of the mF factor omitted from the simple MQH correlation.

360

COMPARTMENT FIRES

MQH correlation for discrete floor-centered fires [21]

Figure 11.14

11.6.2

Fully developed fires

The fully developed fire is defined as the state where all of the fuel that can get involved is involved. It does not necessarily mean that all of the fuel is burning, since the lack of air would prevent this. In most building occupancies, the fuel load is high enough to lead to a significant fire that can threaten the structure. This means a significantly high temperature is produced for a long time. Hence, many fully developed fire studies have been motivated to establish the thermal characteristics needed to assess the structural fire resistance. Consequently, these studies have aimed to establish correlations for gas temperature and the fuel burning rate (or more exactly, its mass loss rate). The latter can

1000

1/2

m /ρ g A H F

o

1/2 o

= 0.01

Vent limit effect

800 Floating flames

∆T °C

o

600

.

mF/ρog1/2AoHo1/2 = 0.04

400

Wall Lining Fires 200 0

Full-scale 1/3-scale

0

0.5

1

Q*

O

Figure 11.15

2/3

1.5 Q*w–1/3

2

Temperature for wall lining fires [25]

CORRELATIONS 1000 Wall and Ceiling

.

Lining Fires

1/2

m /ρ g A H F

o

o

1/2 o

361

= 0.01

∆T °C

800 600

.

mF/ρog1/2AoHo1/2 = 0.04

400 200 0

Full-scale 1/3-scale

0

0.5 Q*

O

Figure 11.16

1 1.5 Q*w–1/3

2

2/3

Temperatures for wall and ceiling lining fires [25]

then give the fire duration for a given fuel mass loading in the occupancy. Occupancy studies have been performed from time to time to give typical fuel loadings (usually in wood equivalent), which may range from 10 to 50 kg=m2 (floor area). One aspect of fully developed fires that will not be addressed here is their production of combustion products. When compartment fires become ventilation-limited, they burn incompletely, and can spread incomplete products such as CO, soot and other hydrocarbons throughout the building. It is well established that the yield of these incomplete products goes up as the equivalence ratio approaches and exceeds 1. More information on this issue can be found in the literature [1]. Fully developed fire studies have been performed over a range of fuel loadings and ventilation conditions, but primarily at scales smaller than for normal rooms. Also the fuels have been idealized as wood cribs or liquid or plastic pool fires. The results have not been fully generalized. The strength of the dimensionless theoretical implication of Equation (11.38) suggests that, for a given fuel, the fully developed, ventilation-limited fire should have dependences as T ¼ function Qw ; mF ; G To

ð11:40Þ

where Qo has been eliminated since Q_ depends only on the air flow rate from Equation (11.23), and that depends on the temperature from the flow equation (e.g. Equation (11.5)). Thus, for a given burning configuration (represented in terms of G), T

hk A m_ F pﬃﬃﬃ pﬃﬃﬃﬃﬃﬃ ; pﬃﬃﬃ pﬃﬃﬃﬃﬃﬃ 1 cp gAo Ho 1 gAo Ho

ð11:41Þ

From Equation (11.21) and recognizing that the heat flux to the fuel depends on the flame and compartment temperatures, it follows that 00

AF q_ ðTÞ m_ F L

ð11:42aÞ

362

COMPARTMENT FIRES

or mF

m_ F pﬃﬃﬃ pﬃﬃﬃﬃﬃﬃ 1 gAo Ho

! 00 4 AF T1 q_ ðTÞ p ﬃﬃﬃﬃﬃ ﬃ pﬃﬃﬃ 4 T1 1 g Ao H o L

ð11:42bÞ

By rearrangement of these functional dependencies, it follows for a given fuel ðLÞ that T and

A AF m_ F pﬃﬃﬃ pﬃﬃﬃﬃﬃﬃ pﬃﬃﬃﬃﬃﬃ ; 1 g Ao H o Ao H o A

ð11:43Þ

Investigators have developed correlations for experimental data in this form. Law developed a correlation for the extensive Conseil International du Baˆ timent (CIB) test series [17] involving wood cribs that covered most of the compartment floor [26]. The test series also addressed the effect of compartment shape by varying its width (W) and depth (D). The correlation for the maximum temperature was given as Tmax

1 e0:1 p ﬃﬃﬃ ﬃ ¼ 6000 ð CÞ

¼

A pﬃﬃﬃﬃﬃﬃ Ao Ho

in m1=2

ð11:44Þ

Most of the CIB tests involved crib arrangements having AF =A of mostly 0.75 [2]; therefore, leaving this parameter out of the correlation may be justified. However, for low fuel loadings, Law recommended that the maximum temperature be reduced accordingly: T ¼ Tmax 1 e0:05C ð CÞ;

mF C ¼ pﬃﬃﬃﬃﬃﬃﬃﬃ AAo

in kg=m2

ð11:45Þ

where mF is the fuel load in kg. The mass loss rate is correlated as pﬃﬃﬃﬃﬃﬃ W _mF ¼ 0:18Ao Ho ð1 e0:036O Þ in kg=s D 1=2 D m_ F pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ < 60 kg=s m5=2 for Ao ðHo Þ W

ð11:46Þ

The data corresponding to these correlations are shown in Figures 11.17 and 11.19. The CIB tests consisted of compartments with heights ranging from 0.5, 1.0 and 1.5 m with dimension ratios of D=H; W=H; H=H coded as data sets: 211, 121, 221 and 441. A total of 321 experiments were conducted in still air conditions. The wood crib fuel loading ranged from 10 to 40 kg/m2 with stick spacing to stick thickness ratios of 13, 1 and 3. The data are plotted in Figures 11.17 and 11.19. The compartment surface area, A, excludes the floor area in the plot variables. An alternative plotting format was used by Bullen and Thomas [6] for the mass loss rate, which shows the effect of fuel type in Figure 11.18. These data include an extensive compilation by Harmathy [27] for wood crib fuels (including the CIB data). Here the fuel area is included, but the compartment area is omitted. This shows the lack of

CORRELATIONS

Figure 11.17

363

Normalized mass loss rate for CIB data [17]

completeness of the correlations. However, they are still invaluable for making accurate estimates for fully developed fires. Harmathy gives a fit to the wood crib data in Figure 11.18 as 8 pﬃﬃﬃ pﬃﬃﬃﬃﬃﬃ 1 gAo Ho > > < ð0:0062 kg=m2 sÞAF ; 0:263 kg=s ð11:47Þ m_ F ¼ pﬃﬃAﬃ F pﬃﬃﬃﬃﬃﬃ pﬃﬃﬃﬃﬃﬃ 1 gAo Ho > > : 0:02631 pﬃﬃgﬃAo Ho ; < 0:263 kg=s AF The latter forﬃ large pﬃﬃﬃﬃﬃﬃ equation corresponds to the CIB data and Law’s correlationpﬃﬃﬃﬃﬃ A=ðAo Ho Þ, i.e. ventilation-limited fires. At the asymptote for large A=ðAo Ho Þ from Figure 11.18, the CIB results are for D ¼ W, roughly pﬃﬃﬃﬃﬃﬃ m_ F 0:13Ao Ho kg=s and for Harmathy with 1 ¼ 1:16 kg=m3 , pﬃﬃﬃﬃﬃﬃ m_ F ¼ 0:086 Ao Ho kg=s with Ao in m2 and Ho in m. These differences show the level of uncertainty in the results.

364

COMPARTMENT FIRES

Figure 11.18

Figure 11.19

Fuel mass loss rate in fully developed fires [6]

CIB compartment fully developed temperature [17]

SEMENOV DIAGRAMS, FLASHOVER AND INSTABILITIES

365

11.7 Semenov Diagrams, Flashover and Instabilities The onset of flashover as considered here is induced by thermal effects associated with the fuel, its configuration, the ignition source and the thermal feedback of the compartment. Several fire growth scenarios will be considered in terms of Semenov diagrams (Chapters 4 and 9) portraying the firepower Q_ and the loss rates Q_ l as a function of compartment temperature. According to Equation (11.31), a ‘thermal runaway’ will occur at the critical temperature where Q_ ¼ Q_ l and the two curves are tangent. The developing fire occurs for < 1 and the result of a critical event will lead to a fully developed fire with > 1 likely. Four scenarios will be examined: 1. Fixed fire area, representative of a fully involved fuel package, e.g. chair, crib, pool fire. 2. Ignition of a second item, representing an initial fire (e.g. chair, liquid spill) and the prospect of ignition of an adjoining item. 3. Opposed flow flame spread, representing spread on a horizontal surface, e.g. floor, large chair, mattress, etc. 4. Concurrent flow flame spread, representing vertical or ceiling spread, e.g. combustible linings. In all of these scenarios, up to the critical condition (flashover), <1

and

00 Q_ ¼ m_ F AF hc

ð11:48Þ

Here AF will be taken to represent the projected external fuel surface area that would experience the direct heating of the compartment. From Equation (11.21), accounting approximately for the effects of oxygen and temperature, and distribution effects in the compartment, the fuel mass flux might be represented for qualitative considerations as 00

00

m_ F ¼ m_ F;o ð1 o Þ þ

T ðT 4 To4 Þ L

ð < 1Þ

ð11:49Þ

00

where m_ F;o refers to the burning flux in normal air. The distribution factors, o and T , are estimated to range from about 0.5 to 1 as the smoke layer descends and mixing causes homogeneous properties in the compartment. The oxygen distribution factor is very approximate and is introduced to account for the oxygen in the flow stream in the lower gas layer. The thermal parameter, T , represents a radiation view factor. For all of these scenarios, the effect of oxygen on the burning rate is expected to be small, as the oxygen concentration does not significantly decrease up to flashover. The loss rate is given as, from Equations (11.31) and (11.32), Q_ l ¼ m_ air cp ½xo þ xw ðtÞ þ xr ðTÞðT To Þ

ð11:50Þ

The loss rate is nearly linear in T for low temperatures, and can depend on time, t, as well. Since T is a function of time, the normal independent variable t should be

366

COMPARTMENT FIRES

Figure 11.20

Fixed area fire

considered as dependent since the inverse function t ¼ tðTÞ should be available, in principle. This functional dependence will not be made explicit in the following.

11.7.1

Fixed area fire

Figure 11.20 shows the behavior of the energy release and losses as a function of compartment temperature (Semenov diagram) for fuels characteristic of cribs and ‘pool’ fires (representative of low flame absorbing surface fuels, e.g. walls). As theﬃ loss curve pﬃﬃﬃﬃﬃ decreases due to increasing time or increasing ventilation factor ðA=Ao Ho Þ, a critical condition (C) can occur. The upward curvature and magnitude of the energy release rate increases with AF hc =L. Thomas et al. [4] has estimated the critical temperatures to range from about 300 to 600 C depending on the values of L. Thus, liquid fuels favor the lower temperature, and charring solids favor the higher. An empirical value of 500 C (corresponding to 20 kW/m2 blackbody flux) is commonly used as a criterion for flashover. The type of flashover depicted here is solely due to a fully involved single burning item. A steady state condition that can be attained after flashover is the ventilation-limited (VL) branch state shown in Figure 11.20 in which Q_ ¼ m_ o hair for ¼ 1. It can be shown that the state of oxygen at the critical condition is always relatively high, with < 1. This clearly shows that oxygen depletion has no strong effect. Also, Thomas et al. [4] show that the burning rate enhancement due to thermal feedback at the critical condition is only about 1.5, at most.

11.7.2

Second item ignition

Figure 11.21 depicts the behavior for a neighboring item when becoming involved due to an initiating first fire. The first fire could be a fixed gas burner as in a test or a fully

SEMENOV DIAGRAMS, FLASHOVER AND INSTABILITIES

Figure 11.21

367

Second item ignited

involved fuel, AF;1 . For illustrative purposes, assume the second item has similar properties to AF;1 , and AF represents the combined area after ignition. Ignition of the second item is possible if the material receives a heat flux higher than its critical flux for 00 ignition. It can possibly receive heat by flame heating ðq_ f Þ and heating from the compartment. The critical condition for ignition occurs at a balance between the compartment heating and the objects heat loss estimated by 00

q_ f þ T ðT 4 To4 Þ ¼ ðTig4 To4 Þ þ hc ðTig To Þ

ð11:51Þ

For remote ignition under autoignition conditions, direct flame heating is small or zero, so the critical compartment temperature needs to be greater than or equal to the autoignition temperature, e.g. about 400–600 C for common solid fuels. For ignition by flame contact (to a wall or ceiling), the flame heat flux will increase with DF , the flame thickness of the igniter, estimated by 00

q_ f 1 expðDF Þ 30 kW=m2 ; usually

ð11:52Þ

The exposed area of the second item exposed ðAF;2 Þ depends on the flame length ðzf Þ and its diameter. For the second item ignition to lead to flashover, the area involved must equal or exceed the total critical area needed for the second item. The time for ignition depends inversely on the exposure heat flux (Equation (11.51)). Figure 11.21 shows the behavior for ignition of the second item, where AF;1 is the fixed area of the first item and AF;C is the critical area needed. The energy release rate of both fuels controls the size of the ‘jump’ at criticality and depends directly on AF hc =L. No flashover will occur if the ‘jump’ in energy for the second item is not sufficient to reach the critical area of fuel, AF;C . The time to achieve the jump or to attain flashover is directly related to the fuel property,

368

COMPARTMENT FIRES

kcðTig To Þ2 . For flame contact, the piloted ignition temperature applies and the ignition time is expected to be small ð 10--100 sÞ.

11.7.3

Spreading fires

The flame spread rate can be represented in terms of the pyrolysis front ðxp Þ from Equation (8.7a): v¼

dxp f dt tig

ð11:53Þ

where f is a flame heating length. In opposed flow spread into an ambient flow of speed uo , this length is due to diffusion and can be estimated by Equation (8.31) as f;opposed ðk=cÞo =uo . This length is small ( 1 mm) under most conditions, and is invariant in a fixed flow field. However, in concurrent spread, this heating length is much larger and will change with time. It can be very dependent on the nature of the ignition source since here f ¼ zf xp . In both cases, the area of the spreading fire (neglecting burnout) grows in terms of the velocity and time as ðvtÞn , with n ranging from 1 to 2. The surface flame spread cases are depicted in Figure 11.22. In opposed flow spread, the speed depends on the ignition time, which decreases as the compartment heats the fuel surface. After a long time, Tsurface ¼ Tig and the growth curves approach a vertical asymptote. Consider the initial area ignited as AF;i and AF;C as the critical area needed for a fixed area fire. The area for a spreading fire depends on the time and therefore T as well. For low-density, fast-responding materials, the critical area can be achieved at a compartment temperature as low as the ignition temperature of the material.

Figure 11.22

Surface spread fire growth

REFERENCES

369

For concurrent spread, the growth rate can be much faster, and therefore the critical condition can be reached at lower compartment temperatures. The dependence of the concurrent flame spread area on both Q_ and the surface temperature of the material make this spread mode very feedback sensitive. The Semenov criticality diagrams for fire growth are useful to understand the complex interactions of the fire growth mechanisms with the enclosure effects. These diagrams can be used qualitatively, but might also be the bases of simple quantitative graphical solutions.

References 1. Karlsson, B. and Quintiere, J. G., Enclosure Fire Dynamics, CRC Press, Boca Raton, Florida, 2000. 2. Engineering Guide to Fire Exposures to Structural Elements, Society of Fire Protection Engineers, Bethesda, Maryland, May 2004. 3. Thomas, P. H., Behavior of fires in enclosures – some recent progress, in 14th International Symposium on Combustion, The Combustion Institute, Pittsburgh, Pennsylvania, 1973, pp. 1007–20. 4. Thomas, P. H., Bullen, M. L., Quintiere, J. G. and McCaffrey, B. J., Flashover and instabilities in fire behavior, Combustion and Flame, 1980, 38, 159–71. 5. Thomas, P. H., Fires and flashover in rooms – a simplified theory, Fire Safety J., 1980, 3, 67–76. 6. Bullen, M. L. and Thomas, P. H., Compartment fires with non-cellulosic fuels, in 17th International Symposium on Combustion, The Combustion Institute, Pittsburgh, Pennsylvania, 1979, pp. 1139–48. 7. Tewarson, A., Some observations in experimental fires and enclosures, Part II: ethyl alcohol and paraffin oil, Combustion and Flame, 1972, 19, 363–71. 8. Kim, K. I., Ohtani, H. and Uehara, Y., Experimental study on oscillating behavior in a smallscale compartment fire, Fire Safety J., 1993, 20, 377–84. 9. Takeda, H. and Akita, K., Critical phenomenon in compartment fires with liquid fuels, in 18th International Symposium on Combustion, The Combustion Institute, Pittsburgh, Pennsylvania, 1981, 519–27. 10. Quintiere, J. G., McCaffrey, B. J. and Rinkinen, W., Visualization of room fire induced smoke movement in a corridor, Fire and Materials, 1978, 12(1), 18–24. 11. Quintiere, J. G., Steckler, K. D. and Corley, D., An assessment of fire induced flows in compartments, Fire Sci. Technol., 1984, 4(1), 1–14. 12. Emmons, H. W., Vent flows, in The SFPE Handbook of Fire Protection Engineering, 2nd edn (eds P.J. DiNenno et al.), Section 2, Chapter 5, National Fire Protection Association, Quincy, Massachusetts, 1995, pp. 2-40 to 2-49. 13. Epstein, M., Buoyancy-driven exchange flow through small openings in horizontal partitions, J. Heat Transfer, 1988, 110, 885–93. 14. Steckler, K. D., Quintiere, J. G. and Rinkinen, W. J., Flow induced by fire in a room, in 19th International Symposium on Combustion, The Combustion Institute, Pittsburgh, Pennsylvania, 1983. 15. Rockett, J. A., Fire induced flow in an enclosure, Combustion Sci. Technol., 1976 12, 165. 16. Tanaka, T. and Yamada, S., Reduced scale experiments for convective heat transfer in the early stage of fires, Int. J. Eng. Performanced-Based Fire Codes, 1999, 1(3), 196–203. 17. Thomas, P.H. and Heselden, A.J.M., Fully-developed fires in single compartment, a co-operative research programme of the Conseil International du Baˆ timent, Fire Research Note 923, Joint Fire Research Organization, Borehamwood, UK, 1972.

370

COMPARTMENT FIRES

18. Quintiere, J. G. and McCaffrey, B. J., The Burning Rate of Wood and Plastic Cribs in an Enclosure, Vol. 1, NBSIR 80-2054, National Bureau of Standards, Gaithersburg, Maryland, November 1980. 19. Quintiere, J. G., McCaffrey, B. J. and DenBraven, K., Experimental and theoretical analysis of quasi-steady small-scale enclosure fires, in 17th International Symposium on Combustion, The Combustion Institute, Pittsburgh, Pennsylvania, 1979, pp. 1125–37. 20. Naruse, T., Rangwala, A. S., Ringwelski, B. A., Utiskul, Y., Wakatsuki, K. and Quintiere, J. G., Compartment fire behavior under limited ventilation, in Fire and Explosion Hazards, Proceedings of the 4th International Seminar, Fire SERT, University of Ulster, Northern Ireland, 2004, pp. 109–120. 21. McCaffrey, B. J., Quintiere, J. G. and Harkleroad, M. F., Estimating room temperatures and the likelihood of flashover using fire test data correlations, Fire Technology, May 1981, 17(2), 98– 119. 22. Mowrer, F. W. and Williamson, R B., Estimating temperature from fires along walls and in corners, Fire Technology, 1987, 23, 244–65. 23. Azhakesan, M. A., Shields, T. J., Silcock, G. W. H. and Quintiere, J. G., An interrogation of the MQH correlation to describe centre and near corner pool fires, Fire Safety Sciences, Proceedings of the 7th International Symposium, International Association for Fire Safety Science, 2003, pp. 371–82. 24. Karlsson, B., Modeling fire growth on combustible lining materials in enclosures, Report TBVV–1009, Department of Fire Safety Engineering, Lund University, Lund, 1992. 25. Azhakesan, M. A. and Quintiere, J. G., The behavior of lining fires in rooms, in Interflam 2004, Edinburgh, 5–7 July 2004. 26. Law, M., A basis for the design of fire protection of building structures, The Structural Engineer, January 1983, 61A(5). 27. Harmathy, T. Z., A new look at compartment fires, Parts I and II, Fire Technology, 1972, 8, 196– 219, 326–51.

Problems 11.1

A trash fire occurs in an elevator shaft causing smoke to uniformly fill the shaft at an average temperature of 200 C. 12 11 200°C

10 9 8

36 m

20°C

7 6 5 4 3 2 1

Floor numbers

PROBLEMS

371

The only leakage from the shaft can occur to the building that is well ventilated and maintained at a uniform temperature of 20 C. The leak can be approximated as a 2 cm wide vertical slit which runs along the entire 36 m tall shaft. (a) Calculate the mass flow rate of the smoke to the floors. (b) What floors receive smoke? Compute neutral plume height. (c) What is the maximum positive pressure difference between the shaft and the building and where does it occur? 11.2

A 500 kW fire in a compartment causes the smoke layer to reach an average temperature of 400 C. The ambient temperature is 20 C. Air and hot gas flow through a doorway at which the neutral plane is observed to be 1 m above the floor. A narrow slit 1 cm high and 3 m wide is nominally 2 m above the floor.

1 cm 2m 1m

Assume the pressure does not vary over the height of the slit. (a) Calculate the pressure difference across the slit. (b) Calculate the mass flow rate through the slit. The slit flow coefficient is 0.7. 11.3

A fire occurs in a 3 m cubical compartment made of 2 cm thick. Assume steady state heat loss through the concrete whose thermal conductivity is 0.2 W/m2 K. By experiments, it is found that the mass loss rate, m_ , of the fuel depends on the gas temperature rise, T, of the compartment upper smoke layer:

m_ ¼ m_ o þ ðTÞ3=2 where m_ o ¼ 10 g=s; ¼ 103 g=sK 3=2 andT ¼ T T1 . The compartment has a window opening 1 m 1 m. The heat of combustion for the fuel is 20 kJ/g and ambient air is at 20 C. Compute the gas temperature rise, T.

3m 1m

11.4

A room 3 m 4 m 3 m high with an opening in a wall 2 m 2 m contains hexane fuel, which can burn in pools on the floor. The ambient temperature is 25 C. The heat of combustion for hexane is 43.8 kJ/g. The construction material of the room is an insulator which responds quickly, and thus has a constant effective heat loss coefficient, hk ¼ 0:015 kW=m2 K. Use this to compute the overall heat loss to the room.

372

COMPARTMENT FIRES (a) A pool 0.8 m in diameter is ignited and the flame just touches the ceiling. What is the energy release rate of this fire? (b) If this fire entrains 10 times stoichiometric air, what is the equivalence ratio for this room fire? (c) What is the average gas temperature of the hot layer for the fire described in (a)? (d) Flashover is said to commence if the temperature rise of the gas layer is 600 C. At this point, more of the hexane can become involved. What is the energy release rate for this condition? (e) What is the flame extent for the fire in (d)? Assume the diameter of the pool is now 1 m. (f) Based on a room average gas temperature of 625 C, what is the rate of air flow through the opening in g/s? Assume the windowsill opening is above the height of the hot gas layer interface of the room. (g) What is the equivalence ratio for the onset of flashover where the gas temperature is 625 C? (Hint: the heat of combustion per unit mass of air utilized is 3.0 kJ/g.) (h) What is the energy release rate when the fire is just ventilation-limited, i.e. the equivalence ratio is one? Assume the air flow rate is the same as that at the onset of flashover as in (f). (i) What is the fuel production rate when the fire is just ventilation-limited?

11.5

A fire in a ship compartment burns steadily for a period of time. The average smoke layer achieves a temperature of 420 C with the ambient temperature being 20 C. The compartment is constructed of 1 cm thick steel having a thermal conductivity of 10 W/m2 K. Its open doorway hatch is 2.2 m high and 1.5 m wide. The compartment has an interior surface area of 60 m2. The fuel stoichiometric air to fuel mass ratio is 8 and its heat of combustion is 30 kJ/g. (a) Compute the mass burning rate (b) If the temperature does not change when the hatch is being closed, find the width of the hatch opening (the height stays fixed) enough to just cause the fire to be ventilationlimited.

11.6

A fire burns in a compartment under steady conditions. The fuel supply rate ðm_ F Þ is fixed.

T T ∞=20°C m· F , fixed

Its properties are: Heat of combustion = 45 kJ/g Stoichiometric air to fuel mass ratio = 15 g air/g fuel

PROBLEMS

373

The ambient temperature is 20 C and the specific heat of the gases is assumed constant at 1 kJ/kg K. The fuel enters at a temperature of 20 C. For all burning conditions 40 % of the energy released within the compartment is lost to the walls and exterior by heat transfer. (a) The compartment has a door 2 m high, which is ajar with a slit 0.01 m wide. The fire is just at the ventilation limit for this vent opening.

2m 0.01 m

(i) Calculate the exit gas temperature. (ii) Calculate the airflow rate. (iii) Calculate the fuel supply rate. (b) The door is opened to a width of 1 m. Estimate the resulting exit gas temperature. State any assumptions.

2m

1m

11.7

A room in a power plant has a spill of diesel fuel over a 3 m diameter diked area. The compartment is made of 20 cm thick concrete, and the properties are given below. The only opening is a 3 m by 2.5 m high doorway. The dimensions of the compartment are 10 m 30 m 5 m high. Only natural convection conditions prevail. The ambient air temperature is 20 C. Other properties are given below: Concrete: k = 1.0 W/m K ¼ 2000 kg/m3 c ¼ 0.88 kJ/kg K

374

COMPARTMENT FIRES Diesel: Liquid density ¼ 918 kg/m3 Liquid specific heat ¼ 2.1 J/g K Heat of vaporization ¼ 250 J/g Vapor specific heat ¼ 1.66 J/g K Boiling temperature ¼ 250 C Stoichiometric air to fuel mass ratio ¼ 15 Heat of combustion ¼ 44.4 kJ/g The diesel fuel is ignited and spreads rapidly over the surface, reaching steady burning almost instantly. At this initial condition, compute the following: (a) The energy release rate. (b) The flame extent. (c) The temperature directly above the fuel spill. (d) The maximum ceiling jet temperature 4 m from the center of the spill. At 100 s, the burning rate has not significantly changed, and the compartment has reached a quasi-steady condition with countercurrent flow at the doorway. At this new time, compute the following: (e) The average smoke layer temperature. (f) The air flow rate through the doorway. (g) The equivalence ratio, . By 400 s, the fuel spill ‘feels’ the effect of the heated compartment. At this time the fuel surface receives all the heat flux by radiation from the smoke layer, T 4, over half of its area. For this condition, compute the following: (h) The energy release rate as a function of the smoke layer temperature, T. (i) Compute the total rate of losses (heat and enthalpy) as a function of the compartment smoke layer temperature. Assume the heat loss rate per unit area q can ﬃﬃﬃﬃﬃbe estimated by conduction only into the concrete from the gas temperature; i.e. kc t is the concrete conductance, where t is taken as 400 seconds. (j) Plot (h) and (i) to compute the compartment gas temperature. Is the state ventilationlimited?

11.8

Compute the average steady state temperature on a floor due to the following fires in the World Trade Center towers. The building is 63.5 m on a side and a floor is 2.4 m high. The center core is not burning and remains at the ambient temperature of 20 C. The core dimensions are 24 m 42 m. For each fuel fire, compute the duration of the fire and the mass flow rate excess fuel or air released from the compartment space. The heat loss from the compartment fire should be based on an overall heat transfer coefficient of 20 W/m2 K with a sink temperature of 20 C. The vents are the broken windows and damage openings and these are estimated to be one-quarter of the building perimeter with a height of 2.4 m.

PROBLEMS 1=2

The airflow rate through these vents is given as 0:5Ao Ho dimensions in m.

375

kg=s with the geometric

(a) 10 000 gallons of aviation fuel is spilled uniformly over a floor and ignited. Assume the fuel has properties of heptane. Heptane burns at its maximum rate of 70 g/m2 s. (b) After the aviation fuel is depleted, the furnishings burn. They can be treated as wood. The wood burns according to the solid curve in Figure 11.17. Assume the shape effect W1 =W2 ¼ 1, A is the floor area and AoHo1/2 refers to the ventilation factor involving the flow area of the vents where air enters and the height of the vents. The fuel loading is 50 kg/m2 of floor area. 11.9

Qualitatively sketch the loss rate and gain rate curves for these three fire states: (a) flame just touching the ceiling, (b) the onset of flashover and (c) the stoichiometric state where the fire is just ventilation-limited. Assume the L curve stays fixed in time and is nearly linear. 1500

G

L D

1

C

1000 G, L (kW) B 50 0

A

0 0 1000

200

400

600

800

Compartment Gas Temperature (°C)

Use the graph above in which the generation rate of energy for the compartment is given by the G curve as a function of the compartment temperature and the loss rate of energy; heat from the compartment is described below. (a) L1 represents the loss curve shortly after the fire begins in the compartment. What is the compartment temperature and energy release rate? Identify this point on the above figure. (b) As the compartment walls heat, the losses decrease and the L curve remains linear. What is the gas temperature just before and after the fire becomes ventilation-limited? Draw the curve and label the points VL and VLþ respectively. (c) Firefighters arrive after flashover and add water to the fire, increasing the losses. The loss curve remains linear. What is the gas temperature just before and after the suppression dramatically reduces the fire? Draw this L curve on the above figure. Assume the loss rate to suppression is proportional to the difference between the compartment gas temperature and ambient temperature, T T1 .

11.10

COMPARTMENT FIRES Identify the seven elements in the graph with the best statement below: 2 Compartment Gas Energy Rates kW

376

4 3

3 5

6 1

2 1

0

7 4

100

200

300

400

500

600

700

Gas Temperature °C

Heat loss rate

_____________

Wood crib firepower

_____________

Unstable state

_____________

Fully developed fire

_____________

Firepower under ventilation-limited conditions

_____________

Flashover

_____________

Critical state

_____________

Stable state

_____________

Heat and enthalpy flow rate

_____________

Firepower of a burning wall

_____________

800

12 Scaling and Dimensionless Groups

12.1 Introduction The Wright brothers could not have succeeded with their first flight without the use of a wind tunnel. Even today, aircraft cannot be designed and developed without wind tunnel testing. While the basic equations of fluid dynamics have been established for nearly 200 years, they still cannot be solved completely on the largest of computers. Wind tunnel testing affords the investigation of flow over complex shapes at smaller geometric scales while maintaining the Reynolds number ðReÞ constant. The Re number represents the ratio of momentum to viscous forces, i.e. u2 l2 ul ¼ ðu=lÞl2 where Newton’s viscosity law has been used (stress ¼ velocity gradient) and is viscosity, is density, u is velocity and l is a length scale. Although the Re is sufficient to insure similarity between the prototype and the model, many phenomena require the preservation of too many groups to ensure complete similarity. Fire falls into that category. However, that does not preclude the use of scale models or dimensionless correlations for fire phenomena to extend the generality to other scales and conditions. This incomplete process is the art of scaling. Partial scaling requires knowledge of the physics for the inclusion and identification of the dominant variables. Many fields use this technique, in particular ship design. There the Froude number, Fr ¼ u2=gl, is preserved at the expense of letting the Re be what it will, as long as it is large enough to make the flow turbulent. As we shall see, Froude modeling in this fashion is very effective in modeling smoke flows and unconfined fire phenomena. Dimensionless variables consisting of the dependent variables and the time and space coordinates will be designated by (^). Properties and other constants that form Fundamentals of Fire Phenomena James G. Quintiere # 2006 John Wiley & Sons, Ltd ISBN: 0-470-09113-4

378

SCALING AND DIMENSIONLESS GROUPS

dimensionless arrangements are commonly termed groups and are frequently designated by ’s. The establishment of the dimensionless variables and groups can lead to the extension of experimental results through approximate formulas. The use of such formulas has sometimes been termed fire modeling as they allow the prediction of phenomena. These formulae are usually given by power laws where the exponents might be established by theory or direct curve-fitting. The former is preferred. These formulas, or empirical correlations, are very valuable, and their accuracy can be superior to the best CFD models. While correlations might be developed from small-scale experiments, they have not necessarily been designed to replicate a larger scale system. Such a process of replication is termed scale modeling. It can prove very powerful for design and accident investigation in fire. Again, the similarity basis will be incomplete, since all the relevant ’s cannot be preserved. Successful scale modeling in fire has been discussed by Heskestad [1], Croce [2], Quintiere, McCaffrey and Kashiwagi [3] and Emori and Saito [4]. Thomas [5,6] and Quintiere [7] have done reviews of the subject. Novel approaches have been introduced using pressure modeling as presented by Alpert [8] and saltwater models as performed by Steckler, Baum and Quintiere [9]. Even fire growth has been modeled successfully by Parker [10]. Some illustrations will be given here, but the interested reader should consult these and related publications in the literature.

12.2 Approaches for Establishing Dimensionless Groups Three methods might be used to obtain nondimensional functionality. First is the Buckingham pi theorem, which starts with a chosen set of relevant variables and parameters pertaining to a specific modeling application. The numbers of physical independent variables are then determined. The number of groups is equal to the number of variables minus the dimensions. Second, the pertaining fundamental partial differential equation is identified and the variables are made dimensionless with suitable normalizing parameters. Third, the governing physics is identified in the simplest, but complete, form. Then dimensionless and dimensional relationships are derived by identities. We will use the last method. An illustration of the three methods is given for the problem of incompressible flow over a flat plate. The flow is steady and two-dimensional, as shown in Figure 12.1, with the approaching constant flow speed designated as u1 .

Figure 12.1

Flow over a flat plate

APPROACHES FOR ESTABLISHING DIMENSIONLESS GROUPS

12.2.1

379

Buckingham pi method

In general, the velocity in the flow ðuÞ is a function of x, y, l, u1 , and . Seven variables and parameters are identified in the problem: , u, u1 , x, y, l and (viscosity). Since three dimensions appear, M (mass), L (length) and T (time), three variables or parameters can be eliminated by forming four dimensionless groups: 1 , 2 , 3 , 4 . Three repeating variables are selected, as u1 , l and , to form the ’s from the other variables. Let 1 ¼ ua1 , lb , c , . Therefore by dimensional analysis 1 must have no dimensions. Equating powers for each dimension gives M: 0¼cþ1 L: 0¼aþbc3 T: 0 ¼ a c

;c ¼ 1 ;a þ b ¼ 2 ;a ¼ 1; b ¼ 1

It is found that 1 ¼ u1 l= Re, the Reynolds number. (Note that ML1 T 1 .) The other ’s can be determined in a like manner, or simply by inspection: 2 ¼ u=u1 ;

3 ¼ x=l

and

4 ¼ y=l

The dimensionless functional relationship follows as x y u ¼ function ; ; Re u1 l l

12.2.2

Partial differential equation (PDE) method

This method starts with knowledge of the governing equation. The governing equation for steady two-dimensional flow with no pressure gradient is @u @u @2u u þv ¼ 2 @x @y @y with boundary conditions: y ¼ 0;

u¼v¼0

y ! 1; u ¼ u1 x ¼ 0;

u ¼ u1

The conservation of mass supplies another needed equation, but it essentially gives similar information for the y component of velocity. The dimensionless variables are selected as: ^ u ¼ u=u1 ; ^x ¼ x=l;

^v ¼ v=u1 ^y ¼ y=l

380

SCALING AND DIMENSIONLESS GROUPS

Substituting gives ^ u

@^ u @^ u þ ^v ¼ @^x @^y

2 @ ^u u1 l @^y2

The same dimensionless functionality is apparent as from the Buckingham pi method.

12.2.3

Dimensional analysis

Let us use a control volume approach for the fluid in the boundary layer, and recognize Newton’s law of viscosity. Where gradients or derivative relationships might apply, only the dimensional form is employed to form a relationship. Moreover, the precise formulation of the control volume momentum equation is not sought, but only its approximate functional form. From Equation (3.34), we write (with the symbol implying a dimensional equality) for a unit depth in the z direction u u2 ðx 1Þ ðx 1Þ y In dimensionless form:

u 2 x u=u1 x u1 l u1 l l y=l

Hence, u ^ u ¼ functionð^x; ^y; ReÞ u1

12.3 Dimensionless Groups from the Conservation Equations The dimensionless groups that apply to fire phenomena will be derived from the third approach using dimensional forms of the conservation equations. The density will be taken as constant, 1 without any loss in generality, except for the buoyancy term. Also, only the vertical momentum equation will be explicitly examined. Normalizing parameters for the variables will be designated as ðÞ . In some cases, the normalizing factors will not be the physical counterpart for the variable, e.g. x=l, where l is a geometric length. The pi groups will be derived and the dimensionless variables will be preserved, e.g. ^ u ¼ u=u and ^x ¼ x=l . After deriving the pi groups fi g, we will then examine their use in correlations, and how to employ them in scale modeling applications. Common use of symbols will be made and therefore they will not always be defined here.

DIMENSIONLESS GROUPS FROM THE CONSERVATION EQUATIONS

Figure 12.2

12.3.1

381

Mass flow rate

Conservation of mass

From Figure 3.5 the mass flow rate can be represented as m_ ¼ uA

ð12:1Þ

as seen from Figure 12.2. The area will be represented in terms of the scale length of A l2 and similarly l3 for volume.

12.3.2

Conservation of momentum

Consider a form of the vertical momentum equation (Equation (3.24)) with a buoyancy term and the pressure as its departure from hydrostatics. In functional form, V

du þ m_ u ð1 ÞgV þ pA þ S dt

ð12:2Þ

as demonstrated in Figure 12.3. The relationship between the terms in Equation (12.2) can be used to establish normalizing parameters. For example, under nonforced flow conditions there is no obvious velocity scale factor (e.g. u1 ). However, one can be determined that is appropriate in natural convection conditions. Let us examine how this is done. Relate the momentum flux term to the buoyancy force: u2 l2 ð1 Þgl3

ð12:3Þ

By the perfect gas law (under constant pressure) 1 T T1 ¼ T1

Figure 12.3

Momentum balance

ð12:4Þ

382

SCALING AND DIMENSIONLESS GROUPS

Hence, it follows that sﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ pﬃﬃﬃﬃ T T1 u gl gl T1

ð12:5Þ

and therefore an appropriate normalizing factor for velocity is pﬃﬃﬃﬃ u ¼ gl

ð12:6Þ

This need not be the only choice, but it is very proper for natural convection and represents an ideal maximum velocity due to buoyancy. Similarly, by equating the unsteady momentum term with buoyancy gives a timescale sﬃﬃﬃ u l ð12:7Þ t g g The pressure and stress can be normalized as p 1 gl

ð12:8Þ

@u u @x l

ð12:9Þ

Using Newton’s viscosity law,

and equating momentum flux and stress terms gives u 2 l l

ð12:10Þ

1 u l momentum shear force

ð12:11Þ

1 u2 l2 with the ratio as the Reynolds number, Re ¼ 1 ¼

Alternatively, writing Equation (12.10) in terms of the dimensionless velocity, and using the first part of Equation (12.5) for u gives (an approximate solution) " ^ u

pﬃﬃﬃﬃ 1 ½ðT T1 Þ=T1 1=2 gll

#1=2

1 Gr

1=4 ð12:12Þ

where Gr denotes the Grashof number. The Gr is an alternative to Re for purely natural convection flows.

12.3.3

Energy equation

It is the energy equation that produces many more groups through the processes of combustion, heat transfer, evaporation, etc. We will examine these processes. Figure 12.4

DIMENSIONLESS GROUPS FROM THE CONSERVATION EQUATIONS

Figure 12.4

383

Energy transfer

is the basis of the physical and chemical processes being considered, again in functional form. The mass flows include fuel (F), oxygen ðO2 Þ, liquid water (l), evaporated water vapor (g) and forced flows (fan). The chemical energy or firepower is designated as Q_ and all of the heat loss rates by q_ . While Figure 12.4 does not necessarily represent a fire in a room, the heat loss formulations of Chapter 11 apply. From Equation (3.48), the functional form of the energy equation is cp V

dT þ m_ cp ðT T1 Þ Q_ q_ m_ w hfg dt

ð12:13Þ

with the last term being the energy rate required for evaporation of water droplets. Recall that the firepower within the control volume is either Q_ ¼ mF hc ;

fuel controlled

ð12:14aÞ

ventilation-limited

ð12:14bÞ

or m_ O2 hc ; Q_ ¼ r

Relating the unsteady energy term (or advected enthalpy rate) with the firepower gives the second group commonly referred to as Q (or the Zukoski number for Professor Edward Zukoski, who popularized its use in fire): Ql 2 ¼

Q_ firepower pﬃﬃﬃ 1 cp T1 g l 5=2 enthalpy flow rate

ð12:15Þ

where here Ql is based on the length scale l. It should be noted that this group gives rise to an inherent length scale for fire in natural convection:

l ¼

2=5 Q_ pﬃﬃﬃ 1 cp T 1 g

ð12:16Þ

This length scale can be associated with flame height and the size of large-scale turbulent eddies in fire plumes. It is a natural length scale for fires.

384

SCALING AND DIMENSIONLESS GROUPS

12.3.4

Heat losses

In Chapter 11 the heat fluxes to solid boundaries of an enclosure were described. They are summarized here for the purpose of producing the relevant groups. Consider first the radiant loss rate through a flow surface, A, of a control volume surrounding the enclosure gas phase in Figure 12.4: 4 4 q_ o ¼ A½"ðT 4 T1 Þ þ ð1 "ÞðTw4 T1 Þ

ð12:17Þ

where " is the gas emissivity and the walls have been assumed as a blackbody. The gas emissivity may be represented as " 1 e l

ð12:18Þ

with being the absorption coefficient, and hence, another is 3 ¼ l

radiation emitted blackbody radiation

ð12:19Þ

This group is difficult to preserve for smoke and fire in scaling, and can be troublesome. An alternate empirical radiation loss to the ambient used for unconfined fires produces the dimensionless group, Xr 4 ¼

q_ o Q_

ð12:20Þ

The heat transfer to the walls or other solid surfaces takes a parallel path from the gas phase as radiation and convection to conduction through the wall thickness, w . This wall heat flow rate can be expressed as q_ w ¼ q_ k ¼ q_ r þ q_ c

ð12:21Þ

The radiation exchange for blackbody walls can be represented as q_ r "ðT 4 Tw4 ÞS

ð12:22Þ

q_ c hc SðT Tw Þ

ð12:23Þ

and convection as

where S is the wall surface area. The heat transfer coefficient, hc , can be explicitly represented in terms of a specific heat transfer correlation, i.e. k hc Ren Pr m l

ð12:24Þ

The conduction loss is q_ k

kw SðTw T1 Þ

ð12:25Þ

DIMENSIONLESS GROUPS FROM THE CONSERVATION EQUATIONS

385

pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ where ðk=cÞw t for a thermally thick wall, and ¼ w for one that is thermally thin. Pi heat transfer groups can now be produced from these relationships. Let us consider these in terms of Q ’s, i.e. Zukoski numbers for heat transfer. Normalizing with the advection of enthalpy, the conduction group is Qk

pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ kw T1 l2 = ðk=cÞw t wall conduction 5 ¼ pﬃﬃﬃ enthalpy flow 1 cp T1 g l 5=2

ð12:26aÞ

Substituting for t from Equation (12.7) gives Qk ¼

ðkcÞ1=2 w 1 cp g1=4 l3=4

ð12:26bÞ

Similarly, for convection, Qc 6 ¼

hc convection pﬃﬃﬃﬃ 1 cp gl enthalpy flow

ð12:27Þ

Qr 7 ¼

3 T1 radiation pﬃﬃﬃﬃ 1 cp gl enthalpy flow

ð12:28Þ

For radiation, we get

In addition, for a wall of finite thickness, w , we must include 8 ¼

w c1=2 g1=4 thickness ¼ w k w l thermal length

ð12:29Þ

It is interesting to estimate the magnitude of these Zukoski numbers for typical wall materials that could be exposed to fire: Qc

Qk 103

and

Qr 102

to

to

102

101

Hence, under some circumstances the heat losses could be neglected compared to Q 101 to 1, the combustion term.

12.3.5

Mass flows

There is a counterpart to the energy Zukoski number for mass flows. These Zukoski mass numbers can easily be shown for the phenomena represented in Figure 12.4 as mFan 9 ¼

fan flow m_ Fan pﬃﬃﬃ 1 g l 5=2 buoyant flow

ð12:30Þ

386

SCALING AND DIMENSIONLESS GROUPS

Another mass flow term could be due to evaporation, e.g. water cooling from a sprinkler or that of gaseous fuel degradation from the condensed phase. From Chapter 9, we can easily deduce that m_ F 10 ¼

m_ F pﬃﬃﬃ 5=2 1 gl

ð12:31Þ

in which mF depends on Gr, Re, Pr, B, o ¼ cp ðTv T1 Þ=L 11 ; ro ¼ YO2 ;1 = ðrYF;o Þ ¼ 12 and other factors involving flame radiation ð3 Þ and suppression. The mass flow rate of vapor due to evaporation can be similarly represented. From Equation (6.35) it is seen that the dimensionless evaporation mass flux is ^_ g00 ¼ mg m

13;b Mg e m_ g00 pﬃﬃﬃﬃ ¼ Qc M e13;v 1 gl

ð12:32Þ

and 13;i ¼

Mg hfg evaporation thermal energy RTi

ð12:33Þ

where b refers to boiling and v the surface vaporizing. The molecular weight ratio is a constant and might be incidental in most cases. Hence, we see that 13 is the material property group that is needed for determining mg .

12.3.6

Liquid droplets

The mass flux arising from the evaporation of liquid droplets is significant to fire scaling applications. Such scaling has been demonstrated by Heskestad [11], and specific results will be discussed later. As a first approximation, independent monodispersed droplets of spherical diameter, Dl , and particle volume density, n000 , can be considered. A conservation of the number of droplets, n, could be represented as a conservation equation in functional form as dn n_ col dt

ð12:34Þ

where n_ col is the loss rate due to droplet collisions with each other and to surfaces. A constitutive equation is needed for this loss, but is too complex to address here. Simply note that there is another group due to this effect: 14 ¼

n_ col t nref

ð12:35aÞ

where the reference droplet number is based on the initial flow rate of liquid water m_ l;o through the nozzle diameter Do at initial droplet size Dl;o . Therefore, m_ l;o ¼ l uo D2o 4

387

DIMENSIONLESS GROUPS FROM THE CONSERVATION EQUATIONS

and nref

m_ l;o t l D3l;o

ð12:35bÞ

n_ col D3l;o V_ l;o

ð12:35cÞ

Then 14 ¼

where V_ l;o is the initial discharge volumetric flow rate. The initial momentum of the jet must also be considered, as it is significant. An effective nozzle diameter can be determined to account for this momentum since sprinkler discharges lose momentum when they strike the droplet dispersal plate. This effective diameter can be found by measuring the initial thrust ðFo Þ of the spray. Consequently, another group follows as 15 ¼

Fo thrust 2 spray momentum _ l Vl;o =Do

ð12:36Þ

This is an important scaling consideration as the spray momentum is a controlling factor in the droplet trajectories. The final consideration for the droplet dynamics comes from the momentum equation applied to a single droplet, as illustrated in Figure 12.5: ml

dul ðul uÞ2 ml g þ CD D2l 4 dt 2

ð12:37aÞ

with the drag force coefficient taken at the small Reynolds number regime, as the droplets are small [11]: CD 20

jul ujDl 1=2

ð12:37bÞ

pﬃﬃﬃﬃ ^ l ¼ Dl =l and ^l ¼ l =1 . The dependent variables are normalized as ^ ul ¼ ul = gl, D

Figure 12.5

Droplet dynamics

388

SCALING AND DIMENSIONLESS GROUPS

Let us examine the droplet motion (Equation (12.37)) along with its evaporation transport (Equation (12.32)) and the conservation of mass for a single drop: l dDl ¼ m_ g00 2 dt

ð12:38Þ

In functional form, and taking the ratio of both sides, this gives 16 ¼

^_ g00 m_ g00 l1=2 m evaporation rate pﬃﬃﬃ ¼ ^ l droplet mass loss rate l Dl g ^l D

ð12:39Þ

From Equations (12.37) and (12.38), the ratio of the last two terms gives ^ 3=2 17 ¼ D

weight of droplet drag force

ð12:40Þ

^ D ^ l Re1=3 , a viscous diameter. where D l This viscous diameter contains the important Re effects that must be addressed (while Re might be ignored in the bulk flow). Hence, let us modify 16 into an alternative group: 1=3

2=3

16 17 ¼ 160

mg Rel ^l

ð12:41Þ

Thus, we can equivalently use either 16 or 160 . Finally, Equation (12.32) gives a specific group for the single droplet in Qc;l as 18 ¼

pﬃﬃﬃﬃ 1 1 Cp gl advection ¼ Qc;l mass transfer hc

ð12:42Þ

where hc is given from Ranz and Marshall [12] as hc D l 1=2 ¼ 0:6 ReDl Pr1=3 k

ð12:43Þ

for 10 ReDl 103. Consequently, 18 can alternatively be expressed as 1=2

^ 18 0 ¼ Pr 2=3 D l

12.3.7

1=2

Rel

ð12:44Þ

Chemical species

A conservation of species with the possibility of chemical reaction can be functionally represented from Equation (3.22) for species i with a chemical yield, yi (mass of species i per mass of reacted fuel): l3

dYi yi Q_ þ m_ Yi ¼ hc dt

ð12:45Þ

DIMENSIONLESS GROUPS FROM THE CONSERVATION EQUATIONS

389

The group that emerges is 19 ¼

yi cp T1 ith enthalpy chemical energy hc

ð12:46Þ

An alternative approach might be to address solely the mixture fraction f (mass of original fuel atoms per mass of mixture) since it has been established that there is a firm relationship between yi and f for a given fuel. Note that f moves from 1 to 0 for the start and end of the fire space and f is governed by Equation (12.45) for yi 0. This then conserves the fuel atoms. Under this approach it is recognized that yi ¼ function ðf Þ

ð12:47Þ

for a given fuel under fire conditions. For the same fuel, the mass fraction varies as Yi l0 .

12.3.8

Heat flux and inconsistencies

In the implementation of scaling or even in the development of empirical correlations, it is almost universally found that radiation heat transfer is not completely included. In addition, the wall boundary thermal properties are rarely included in fire correlations. Let us consider the example of geometric scale modeling in which l is a physical dimension of the system. Then, for geometrically similar systems, we seek pﬃﬃﬃﬃﬃﬃ ﬃ to maintain the same temperature at homologous points ð^x; ^y; ^zÞ and time ^t ¼ t g=l. Figure 12.6 shows the model and prototype systems. Let us examine the consequences of the fire heat flux for various choices in scaling. Note by preserving Q ð2 Þ, it is required that the model firepower be selected according to 5=2 _Qm ¼ Q_ p lm lp

ð12:48Þ

or Q_ l5=2. From Equation (12.13), if all the heat losses also go as l5=2 , then T l0 , or temperature is invariant with scale size. Let us examine the behavior of the wall or surface heat losses under various scaling schemes to see if this can be accomplished. It will be seen that all forms of heat transfer cannot be simultaneously preserved in scaling.

Figure 12.6

Model and prototype geometric similar systems

390

SCALING AND DIMENSIONLESS GROUPS

For complete scaling in model heat transfer, the following must be preserved: ; l1

3 ¼ l ðkcÞ1=2 w 1 cp gl3=4 hc pﬃﬃﬃﬃ 6 ¼ Qc ¼ 1 cp gl 5 ¼ Qk ¼

3 T1 pﬃﬃﬃﬃ 1 cp gl 1=2 1=4 ðkcÞ g

;ðkcÞw l3=2 ;hc l1=2

7 ¼ Qr ¼ 8 ¼

w

kw

l

;T1 l1=6 w

;w

kw l1=4

!

ðkcÞ1=2 w

In addition, for turbulent convection it is commonly found that the Nusselt number is important, Nu hc l=k. Since pﬃﬃﬃ !4=5 k 1 gl3=2 hc l1=5 l (for laminar flow we would have, instead, hc l1=4 ), we see that inconsistencies emerge: hc l1=2 or l1=5 and T1 l1=6 , while it should be l0 as it is difficult to control the ambient temperature. Several strategies can be employed to maintain partial scaling. This is the meaning of the art of scaling. It requires some insight into the importance of competing effects. Let us first consider maintaining 5 and 8 constant. This preserves wall conduction effects. Consequently, recognizing that kw w , usually for materials, and then assuming cw l0 (since specific heats for solids do not vary much among materials), from 5 :

kw w l3=4

from 8 :

w l1=4

and

Conduction heat flux follows: 00

q_

kw l3=4 ðTw T1 Þ 1=4 l1=2 w l

ð12:49Þ

and the wall conduction heat loss rate varies as required, l5=2 . Let us try to preserve convection. If we ignore 6 but maintain the more correct turbulent boundary layer convection (Equation (12.24)), then 00

q_ hc ðT Tw Þ l1=5 l0 l1=5

ð12:50Þ

DIMENSIONLESS GROUPS FROM THE CONSERVATION EQUATIONS

391

Finally, consider radiation. Under optically thin conditions ð3 ¼ l smallÞ, 00

q_ lT 4 l

ð12:51Þ

and for the optically thick case ð l largeÞ, 00

q_ T 4 1 e l

ð12:52Þ

with Equation (12.18). If we could preserve 3 , then 00

q_ l0 ; optically thin

ð12:53aÞ

and 00

q_ l0 ; optically thick

ð12:53bÞ

This preservation requires that l1 . It could be accomplished by changing the fuel in the model to one that would yield more soot. This may not be done with great accuracy, but is feasible. The convection heat flux is usually much smaller than radiation flux in significant fires. Preserving 5 , 6 and 8 yields q_ 00 l1=2 , which is inconsistent with the radiation flux (Equation (12.53)). However, if the same fuel is used in the model as in the prototype and no change occurs in as Yi , then the radiation flux is 00

q_ l; for small and 00

q_ l0 ; for large If the fuel is modified such that l1=2 , then the radiant flux for small becomes consistent with conduction as q_ 00 l1=2 , and the radiation heat loss goes as required, l5=2 . Alternatively, if radiation is believed to dominate, it might be useful to consider large and ignore 5 but require that 00

q_

kw T l0 w

ð12:54Þ

As a consequence of this choice, while still maintaining 8 and therefore w l1=4 , then kw w l1=4 . Hopefully, this discussion has shown some of the issues related to achieving complete scaling. However, thoughtful partial scaling is still a valid approach for obtaining fairly accurate results for complex geometries. Remember, in scaling, the turbulence and combustion phenomena are inherent in the system, and therefore no empirical models are needed as in CFD approaches. In addition to the above inconsistencies, it is also not possible to preserve the Re number throughout (Equation (12.11)). However, if the

392

SCALING AND DIMENSIONLESS GROUPS

pressure of the system is altered in order to change 1 between the model and the prototype, 1 can be preserved [8], i.e. pﬃﬃﬃ 1 gl3=2 Re ¼ pl3=2 In order to preserve the Re number, pm ¼ pp

3=2 lm lp

ð12:55Þ

Another alternative is to conduct a scale model experiment in a centrifuge in which g is now increased [13]. Modifying both p and g in the model can allow the preservation of more groups. Thus, scaling in fire is not complete, but it is still a powerful tool, and there are many ways to explore it. Illustrations will be given later of successful examples in scale modeling and correlations to specific fire phenomena.

12.3.9

Summary

Table 12.1 gives a summary of the dimensionless variables. Two additional groups have been added, the Weber number, We, to account for droplet formation and the Nusselt number, Nu ¼ hc l=k, to account for gas phase convection. A corresponding Nusselt Dimensionless variables and scalling in fire

Table 12.1 Variable/group

Dimensionless

Velocity, u Temperature, T Pressure, p Concentration, Yi Droplet number, n Droplet diameter, Dl 00

_F Burning rate per area, m Coordinates x; y; z Time, t

i

l t pﬃﬃﬃﬃﬃﬃﬃ l=g

u l1=2 T l0 p l Yi l0 n l3=2 12 ! Dl l1=2 00

m_ F

hc l Nu ¼ cp Pr

xi l t l1=2

Pi groups:

1

inertia ; Re viscous

u Dependent: ^u ¼ pﬃﬃﬃﬃ gl T T^ ¼ T1 p ^p ¼ 1 gl Yi Yi;1 n nref Dl l 00 m_ F l Independent: x

Scaling/comment

Re ¼

pﬃﬃ 1 gl3=2

Usually ignored

DIMENSIONLESS GROUPS FROM THE CONSERVATION EQUATIONS Table 12.1

393

ðContinuedÞ

Variable/group firepower ; Q 2 enthalpy rate radiant emission 3 ideal emission radiant loss 4 ; Xr firepower conduction ; Qk 5 enthalpy convection ; Qc 6 enthalpy radiation ; Qr 7 enthalpy thickness 8 thermal length fan flow ; mFan 9 advection fuel flow ; mF 10 advection sensible ; o 11 latent available O2 ; ro 12 stoichiometric O2 evaporation energy 13 sensible energy collision loss 14 initial particles spray thrust 15 jet momentum evaporation rate 16 droplet mass loss weight of droplet ^ ; D 17 drag force advection 18 mass transfer ith enthalpy 19 chemical energy droplet momentum 20 surface tension enthalpy 21 combustion energy convection 22 conduction

Dimensionless

Scaling/comment

Q_ pﬃﬃﬃ 1 cp T gl5=2

Significant in combustion

l

l1 ; gas radiation important

Xr ¼

q_ r Q_

Xr l0 , important for free burning

1=2 ðkcÞw 1 cp g1=4 l3=4

hc pﬃﬃﬃﬃ 1 cp gl 3 T1 pﬃﬃﬃﬃ 1 cp gl c1=2 g1=4 w k w l m_ Fan pﬃﬃﬃ 1 gl5=2

kw w l3=4 , conduction important hc l1=2 , convection important T1 l1=6 , inconsistent with others w l1=4 , thickness of boundaries m_ Fan l5=2 , forced flows

m_ F pﬃﬃﬃ 1 g l5=2

Fuel mass flux depends on B, Gr, Re, etc.

cp ðTv T1 Þ=L

Burning rate term

YO2 ;1 rYF;o Mg hfg RTi ^ncol ¼

Burning rate term ‘Activation’ of vaporization n_ col V_ l;o =D3l;o

Fo

l V_ l;o =Do 00 pﬃ m_ g l pﬃﬃﬃ l Dl g

2

1=2

^ Re Pr 2=3 D l l yi cp T1 hc l u2l Dl We ¼ cp T1 hc =r Nu ¼ hc l=k

Fo l3 ; Do is an effective nozzle diameter, Do l m_ g00 l0 (see 17 )

^ l Re1=3 ^ ¼ D D l 1=2

n_ col l, collision number rate

Dl l1=2 Dl l1=4 , inconsistent with 17 yi l0 Dl l1 , inconsistent with 17 Nearly always constant hc l1

394

SCALING AND DIMENSIONLESS GROUPS

number (or, more precisely, a Sherwood number, Sh ¼ hl=ð1 Di Þ, where Di is a mass diffusion coefficient for species i in the medium) could also be added. These last two additions remind us of the explicit presence of fluid property groups, the Prandtl number, Pr ¼ k=ðcp Þ, and the Schmidt number, Sc ¼ 1 Di =. For media consisting mostly of air and combustion products these property groups are nearly constant and would not necessarily appear in fire correlations. However, for analog media such as smoke to water, they could be important. This set of dimensionless groups and variables represents a fairly complete set of dependent variables and the independent coordinates, time, property and geometric parameters for fire problems. The presentation of dimensionless terms reduces the number of variables to its minimum. In some cases, restrictions (e.g. steady state, twodimensional conditions, etc.) will lead to a further simplification of the set. However, in general, we should consider fire phenomena, with water droplet interactions, that have the functional dependence as follows: n o ^_ 00 ; m ^_ 00 ; ^ ^ l; m ^ u; ^ ul ; T^ ; Y^i ; ^ p; D n ¼ functionf^xi ; ^t; i ; i ¼ 1; 22g g F

ð12:56Þ

together with boundary and initial conditions included. As seen from Table 12.1, it is impossible to maintain consistency for all groups in scaling. However, some may not be so important, and others may be neglected in favor of a more dominant term. This is the art of scaling. In correlations derived from experimental (or computational) data, only the dominant groups will appear.

12.4 Examples of Specific Correlations The literature is filled with empirical formulas for fire phenomena. In many cases, the dimensionless groups upon which the correlation was based have been hidden in favor of engineering utility, and specific dimensions are likely to apply. For example, a popular formula for flame height from an axisymmetric source of diameter D is zf ¼ 0:23Q_ 2=5 1:02D with zf and D in meters and Q_ in kW. However, the more complete formulation from Heskestad [14] is Equation (10.50a): zf scp T1 3=5 ¼ 15:6QD 2=5 1:02 D hc which involves 2 and 21 . In contrast, an alternative, Equation (10.49), is an implicit equation for the flame height, given as zf rcp T1 ¼ function QD ; ; Xr D Yox;1 hc

EXAMPLES OF SPECIFIC CORRELATIONS

395

which includes the effect of free stream oxygen concentration and the flame radiation fraction. These additional effects were included based on theory, and more data need to be established to show their complete importance. The examples presented in Chapter 10 for plumes demonstrate a plethora of equations grounded in dimensionless parameters. Some other examples for plumes will be given here.

12.4.1

Plume interactions with a ceiling

For the geometry in Figure 12.7, the maximum temperature in the ceiling jet has been correlated by Alpert [15]: ðT T1 Þ=T1 2=3

QH

¼f

r H

ð12:57Þ

as seen in the plot found in Figure 12.7. It might not be obvious why the Zukoski number based on the height H has a 2/3 power. This follows from Equations (12.2), (12.5) and

Figure 12.7

Ceiling jet maximum temperature [15]

396

SCALING AND DIMENSIONLESS GROUPS

(12.13), in which l H and u ½½ðT T1 Þ=T1 gH1=2 based on a weakly buoyant 2=3 plume. Consequently, ½ðT T1 Þ=T1 is deduced as QH . Alpert [15] shows similar results for velocity as pﬃﬃﬃﬃﬃﬃﬃ r u= gH ¼f ð12:58Þ 1=3 H QH

12.4.2

Smoke filling in a leaky compartment

Zukoski [16] has shown theoretical results that have stood the test of many experimental checks for the temperature and descent of a uniform smoke layer in a closed, but leaky ðp 0Þ, compartment. Similar to the scaling used for the ceiling jet, he found that t zlayer ¼ function ð12:59Þ t H where pﬃﬃﬃﬃ H S t pﬃﬃﬃ 1=3 2 H gQ H

where S is the floor area. Similarly, the layer temperature follows as t ½ðT T1 Þ=T1 layer ðt=t Þ for QH small: ¼ function 2=3 t 1 z =H layer QH These relationships are shown in Figure 12.8.

Figure 12.8

Compartment smoke filling [16]

ð12:60Þ

EXAMPLES OF SPECIFIC CORRELATIONS

Figure 12.9

12.4.3

397

Burning rate within vertical tubes [17]

Burning rate

The burning rate for simple geometric surfaces burning in a forced flow ðu1 Þ or purely buoyant has been presented in Chapter 9. In general, it might be found that steady burning can be expressed as x m_ F x ¼ function ; Grl ; Rel ; Bm l 00

ð12:61Þ

where Bm is the modified Spalding-B number based on Lm (Equation (9.72)), and primary radiation terms in the form of Xr , l and Qr , plus other effects such as water mg (Equation (9.78)), can be included. Some additional examples of burning rate correlations are shown in Figure 12.9 for burning within vertical tubes [17] and Figure 12.10 for burning vertical cylinders and walls [18]. The latter wall data are subsumed in the work shown in Figure 9.11(b). Here the Grashof number was based on a flame temperature and was varied by pressure: gl3 ½p1 =ðRT1 Þ2 Tf T1 Grl 2 T1

ð12:62Þ

This is an example of pressure modeling done at Factory Mutual Research (FM Global) by deRis, Kanury and Yuen [17] and Alpert [8]. In pressure modeling, Gr is preserved with p1 l3=2 along. Therefore m_ 00F l=, is also preserved and 00

Q

l1 l1=2 l3=2 m_ F l2 3=2 l0 5=2 p1 ð1 T1 Þl l 3

T l0 but radiation is not preserved as l 1 l p1 l l1=2 and Qr p1ﬃl l1=2 l1=2 .

398

SCALING AND DIMENSIONLESS GROUPS

Figure 12.10

12.4.4

Burning rate on vertical cylinders and walls [8]

Compartment fire temperature

In Chapter 11 it was shown that control volume theory for the bulk compartment smoke properties could be expressed in dimensionless solutions. The characteristic length scale involves the geometric components of wall vents as l ¼ ½Ao ðHo Þ1=2 2=5 . Hence, the MQH correlation [18] leads to T T1 2=3 ¼ 1:6Ql Q1=3 w T1

ð12:63Þ

for primarily, centered, well-ventilated, floor fires. Variations of this result have been shown for other fuel geometries (i.e. Equation (11.39)). For the fully developed fire, various correlations have sought to portray the temperature in these fires in order to predict the impact on structures. Chapter 11 highlights the CIB work on wood cribs and the corresponding correlation by Law [19]. It is instructive

EXAMPLES OF SPECIFIC CORRELATIONS

399

to add the correlation by Babrauskas [20] that originated from a numerical solution to describe compartment fires. From his set of solutions, for the steady maximum temperature, he fitted parameters as T T1 ¼ ð1432 CÞ1 2 3 4 5

ð12:64aÞ

From the dimensionless results established in Table 12.1, this result has the dimensionless counterpart as 1 ¼

1 þ 0:51 ln ; 1 0:05 ðln Þ5=3 ;

< 1 for wood cribs: >1

ð12:64bÞ

and pﬃﬃﬃ pﬃﬃﬃﬃ k o 1 g Ao H o ¼ Q1 00 o sm_ F;1 AF where ko ¼ 0:51 and s ¼ stoichiometric air/fuel, or pﬃﬃﬃﬃ Ao H o 0:5hc AF 21 1

Qo 1 ¼ 4 A s T 4 Tboil A 7

ð12:64cÞ

for liquid pool fires. In addition, "

pﬃﬃﬃﬃﬃﬃ 2=3 1=3 # 1=3 Ao H o 2=3 8 2 ¼ 1:0 0:94exp 54 Qo kw 5 A

ð12:64dÞ

"

2=5 # pﬃﬃﬃﬃﬃﬃ 3=5 2=5 Ao Ho t 5=4 3 ¼ 1:0 0:92exp 150 5 ð12:64eÞ Q3=5 o ðkcÞw A 1=3 3=2 4 ¼ 1:0 0:205H 0:3 7

ð12:64fÞ

hc hc;ideal

ð12:64gÞ

and 5 ¼ combustion efficiency

This fairly complete correlation shows that effects of fuel properties and perhaps vent radiation with the appearence of 7 ; as the physics are not evident, while these are missing in the correlation of Law [19] (Equation (11.45)). Yet Law’s correlation includes the effect of a compartment shape. Also in Equation (12.64), the dimensionless groups are indicated compared to the dimensional presentation of Babrauskas, in units of kJ, kg, m, K, s, etc.

400

SCALING AND DIMENSIONLESS GROUPS

Effect of water spray on a pool fire

Figure 12.11

12.4.5

Effect of water sprays on fire

Heskestad [21] has shown that droplet sprays can be partially scaled in fire. He developed a correlation of the water flow rate ðV_ l;o Þ needed for extinction of a pool fire, as shown in Figure 12.11. Heskesrtad determines that the water flow needed for extinction is 0:41 1:08 ml V_ l;o ½HðmÞ0:4 Q_ ðkWÞ ¼ 5:2 Do;e ðmmÞ s

ð12:65aÞ

This followed from the data plot in Figure 12.12. The dimensionless rationale for this correlation is that, at extinction for nozzle overhead acting on an open fire, the important groups for geometric similarity are 2 , 9 and 15 , such that we obtain the following corresponding relationship:

l V_ l;o pﬃﬃﬃ 5=2 1 gl

¼ function

Extinction

Q_ Do pﬃﬃﬃ 5=2 ; 1 cp T1 gl l

ð12:65bÞ

Here, Heskestad selects the nozzle diameter based on satisfying 15 , i.e. the effective nozzle diameter (not its actual, due to the complex discharge configurations) is found from

Do;e ¼

2 4 l V_ l;o Fo

!1=2 ð12:66Þ

Here l is taken as H, the height of the nozzle above the fire. These data are shown to be representative for fires of about 1 to 103 kW, H 0:1 to 1 m, V_ l;o 1 to 50 ml/s and fire diameter, D 0:1 to 1.0 m. In this spray scaling, the droplet size was also a consideration, with Heskestad attempting to maintain 17 constant with Dl;o l1=2 and l varied at 1 and 10 (or D 0:1 and 1 m). Hence, the nozzles used were similar by preserving 17 . As a

SCALE MODELING

Figure 12.12

401

Critical water flow rate at extinction [21]

consequence, 16 gives m_ 00g l0 for the droplet evaporation flux, and the droplet lifetime then scales as tEvap ¼

ml D3 0 l 2 Dl l1=2 m_ g l D

ð12:67Þ

This complies with the general timescale. While this scaling is not perfect, as Rel and 18 are not maintained, the correlation for similar nozzles is very powerful. Moreover, the basis for scale modeling fires with suppression is rational and feasible, and small-scale design testing can be done with good confidence.

12.5 Scale Modeling The use of scale modeling is a very powerful design and analysis tool. It has been used by various fields as demonstrated in the first International Symposium on Scale Modeling [22] and those following. This nearly abandoned art brought together researchers examining structures, acoustics, tidal flows, aircraft, ship, vehicle design, fire and more. While computational methods have taken the forefront today, scale models can

402

SCALING AND DIMENSIONLESS GROUPS

Figure 12.13

Gas temperatures in an enclosure [2]

still offer the possibility of more accuracy as well as a platform to validate these computer solutions. Some specific examples for fire will be considered for illustration. All of these are based on partial scaling with Re ignored (but maintained large enough for turbulent flow to prevail), and gas phase radiation is not generally preserved.

12.5.1

Froude modeling

The term ‘Froude modeling’ has been coined (probably from Factory Mutual researchers) as this is scaling fire by principally maintaining Q constant. It is called Froude modeling as that group pertains to advection and buoyancy and requires that the velocities must scale as u ðglÞ1=2 from Equation (12.6). Heskestad [1] and Croce [2] have used this effectively. They show that wood cribs burning in enclosures can produce scaling that obeys ( pﬃﬃﬃﬃﬃﬃﬃ ) 2 _ l=g h c w2 m F 8 6 c w T^ ; Y^i ; ¼ ; 8 ¼ ¼ function ^xi ; ^t; mF;o ; P; ð12:68Þ k w tF tF m_ F;o 5 kw where P is the crib porosity from Equation (9.83), m_ F;o is the free-burn rate of the crib and tF is the burn time from mF =m_ F;o . The results for maximum burning conditions over an ensemble of configurations and scales are shown in Figures 12.13 and 12.14. In these experiments, similar cribs and geometrically similar enclosures were used at the three geometric scales, such that Q_ l5=2 , and the wall materialpproperties were changed with ﬃﬃﬃﬃﬃﬃ the scale as required. These results show the effect of Ao Ho =m_ F;o (or the air flow rate over the fuel supply rate) and the vent height ratio, Ho =H. Froude modeling was also used to prove that a code-compliant design for atrium smoke control was faulty, and consequently the code was blamed for smoke damage to a

SCALE MODELING

Figure 12.14

403

CO and CO2 near ceiling gas concentrations [2]

large compartment store when a scale model of the conditions was introduced into a civil court case [23]. Figure 12.15 shows aspects of this scale model testing. The fire occurred in the Santa and sled that caused it to fall to the basement of the atrium of the store. By considering the fan and fire effects in the scaling, the smoke-control system was simulated. The recorded results demonstrated that the high-velocity input of the makeup air destroyed the stratification of the smoke layer, and smoke was pushed throughout the department store from the atrium. Reviewing Table 12.1, it is seen that several groups are not preserved in Froude modeling. The Re ð1 Þ is ignored, but the scale must be large enough to insure turbulent flow ðRe > 105 Þ. This is justified away from solid boundaries where Re is large and inviscid flow is approached. In the Heskestad–Croce application only two conduction heat transfer groups are used: 5 and 8 . The groups in Equation (12.68) were maintained in the scaling by varying the wall boundary material properties and its thickness. Alternatively, one might satisfy convection near a boundary by invoking 6 and 8 where the heat transfer coefficient is taken from an appropriate correlation involving Re (e.g. Equation (12.38)). Radiation can still be a problem because re-radiation, 7 , and flame (or smoke) radiation, 3 , are not preserved. Thus, we have the ‘art’ of scaling. Terms can be neglected when their effect is small. The proof is in the scaled resultant verification. An advantage of scale modeling is that it will still follow nature, and mathematical attempts to simulate turbulence or soot radiation are unnecessary.

12.5.2

Analog scaling methods

It has been common to use two fluids of different densities to simulate smoke movement from fire. One frequently used approach is to use an inverted model in a tank of fresh water with dyed or colored salt water, for visualization, in order to simulate the hot

404

SCALING AND DIMENSIONLESS GROUPS

Figure 12.15

(a) Fire occurred on the sled. (b) Scale model of the atrium. (c) Recording smoke dynamics in the scale model

405

SCALE MODELING

Figure 12.16

Plume front rise-time simulated by saltwater modeling [24,25]

combustion products. This follows from Equation (12.13) for combustion and Equation (12.45) for the saltwater flow rate as a source m_ s . We write the energy equation as T T1 3 d cp ðT T1 Þ=T1 þ m_ cp 1 T1 l Q_ ð12:69aÞ dt T1 for an adiabatic system and the salt species concentration equation as s l3

dys þ m_ Ys m_ s dt

ð12:69bÞ

The initial condition is that ðT T1 Þ=T1 ¼ Ys ¼ 0 and at solid boundaries ð@T=@n ¼ @Ys =@n ¼ 0, with no heat or mass flow. Of course, radiation is ignored, but this is justified far away from the source. Note that the dimensionless source terms are Q and ms where t ¼ ðl=gÞ1=2 and l is chosen according to a geometric dimension, or as Equation (12.16) for an unconfined plume depending on whether the flow is confined or unconfined. Figure 12.16 shows the rising fire plume (or falling salt plume) of Tanaka, Fujita and Yamaguchi [24] from Chapter 10 compared to the saltwater data of Strege [25]. The saltwater data correspond to the long-time regime, and follow the combustion data from Tanaka.

Figure 12.17 Saltwater smoke simulation in a two-story compartment (left) model (inverted) [26]

406

SCALING AND DIMENSIONLESS GROUPS

Figure 12.18

Dimensionless smoke front arrival times at various locations in the two-story compartment [26]

Another example is illustrated in Figure 12.17 for a compartment with an open first floor door and ceiling vents on both floors. This could produce a complex flow pattern for smoke from a fire. Event times were recorded, for a fire originating on the first floor as derived in dimensionless form from the saltwater model results, and compared with results from the Fire Dynamics Simulator (FDS) code from NIST. Figure 12.18 shows a comparison between event times in this two-room system for the saltwater scaled times and the corresponding time for a combustion system as computed using FDS by Kelly [26], where l was based on the height of the rooms. The FDS computations varied the fire from 101 to 103 kW compared to the very small salt source. It is seen that the fluid motion is simulated very well over this range of firepower, as the event times are independent of the dimensionless firepower. Relating the fire temperatures to the saltwater concentration is problematic, but can be done by careful measurements. Yao and Marshall [27] made salt concentration measurements by using a nonintrusive laser fluorescence technique. These have been related to the temperature distribution of a ceiling jet. Figure 12.19 shows the laser signal distribution for salt concentration in the impinging region of a plume and its ceiling jet. The saltwater, low-buoyancy system has a tendency to become laminar as the ceiling jet propagates away from the plume due to the buoyancy suppressing the turbulence.

Figure 12.19

Fluorescent saltwater concentration distribution in a ceiling jet [27]

REFERENCES

Figure 12.20

407

Ceiling jet temperature distribution [27]

Figure 12.20 shows the ceiling jet temperature distribution normalized by the maximum rise in the jet as a function of distance from the ceiling normalized with the jet thickness. The results apply to salt concentrations or temperature, and show that the distribution is independent of the distance from the plume. This saltwater analog technique is a powerful method for examining fire-induced flows, in particular with the technology demonstrated by Yao and Marshall [27].

References 1. Heskestad, G., Modeling of enclosure fires, Proc. Comb. Inst., 1973, 14, pp. 1021–30. 2. Croce, P. A., Modeling of vented enclosure fires, Part 1. Quasi-steady wood-crib source fires, FMRCJ.I.7A0R5.G0, Factory Mutual Research, Norwood, Massachusetts, July 1978. 3. Quintiere, J. G., McCaffrey, B. J. and Kashiwagi, T., A scaling study of a corridor subject to a room fire, Combust. Sci. Technol., 1978, 18, 1–19. 4. Emori, R. I. and Saito, K., A study of scaling laws in pool and crib fires, Combust. Sci. Technol., 1983, 31, 217–31. 5. Thomas, P. H., Modeling of compartment fires, Fire Safety J., 1983, 5, 181–90. 6. Thomas, P. H., Dimensional analysis: a magic art in fire research, Fire Safety J., 2000, 34, 111–41. 7. Quintiere, J. G., Scaling applications in fire research, Fire Safety J., 1989, 15, 3–29. 8. Alpert, R. L., Pressure modeling of fires controlled by radiation, Proc. Comb. Inst., 1977, 16, pp. 1489–500. 9. Steckler, K. D., Baum, H. R. and Quintiere, J. G., Salt water modeling of fire induce flows in multi-compartment enclosures, Proc. Comb. Inst., 1986, 21, pp. 143–9. 10. Parker, W. J. An assessment of correlations between laboratory experiments for the FAA Aircraft Fire Safety Program, Part 6: reduced-scale modeling of compartment at atmospheric pressure, NBS-GCR 83 448, National Bureau of Standards, February 1985. 11. Heskestad, G., Scaling the interaction of water sprays and flames, Fire Safety J., September 2002, 37, 535–614.

408

SCALING AND DIMENSIONLESS GROUPS

12. Ranz, W. E. and Marshall, W. R., Evaporation from drops: Part 2, Chem. Eng. Prog., 1952, 48, 173. 13. Altenkirch, R. A., Eichhorn, R. and Shang, P. C., Buoyancy effects on flames spreading down thermally thin fuels, Combustion and Flame, 1980, 37, 71–83. 14. Heskestad, G., Luminous heights of turbulent diffusion flames, Fire Safety J., 1983, 5, 103–8. 15. Alpert, R. L., Turbulent ceiling-jet induced by large-scale fires, Combust. Sci. Technol., 1975, 11, 197–213. 16. Zukoski, E. E., Development of a stratified ceiling layer in the early stages of a closed-room fire, Fire and Materials, 1978, 2, 54–62. 17. deRis, J., Kanury, A. M. and Yuen, M. C., Pressure modeling of polymers, Proc. Comb. Inst., 1973, 14, pp. 1033–43. 18. McCaffrey, B. J., Quintiere, J. G. and Harkleroad, M. F., Estimating room temperatures and the likelihood of flashover using fire test data correlations, Fire Technology, May 1981, 17, 98–119. 19. Law, M., A basis for the design of fire protection of building structures, The Structural Engineer, January 1983, 61A. 20. Babrauskas, V., A closed-form approximation for post-flashover compartment fire temperatures, Fire Safety J., 1981, 4, 63–73. 21. Heskestad, G., Extinction of gas and liquid pool fires with water sprays, Fire Safety J., June 2003, 38, 301–17. 22. Emori, R. I. (ed.), International Symposium on Scale Modeling, The Japan Society of Mechanical Engineers, Tokyo, Japan, 18–22 July 1988. 23. Quintiere, J. G. and Dillon, M. E., Scale model reconstruction of fire in an atrium, in Second International Symposium on Scale Modeling (ed. K. Saito), University of Kentucky, Lexington, Kentucky, 23–27 June 1997. 24. Tanaka, T., Fujita, T. and Yamaguchi, J., Investigation into rise time of buoyant fire plume fronts, Int. J. Engng Performance-Based Fire Codes, 2000, 2, 14–25. 25. Strege, S. M., The use of saltwater and computer field modeling techniques to determine plume entrainment within an enclosure for various fire scenarios, including inclined floor fires, MS Thesis, Department of Fire Protection Engineering, University of Maryland, College Park, Maryland, 2000. 26. Kelly, A. A., Examination of smoke movement in a two-story compartment using salt water and computational fluid dynamics modeling, MS Thesis, Department of Fire Protection Engineering, University of Maryland, College Park, Maryland, 2001. 27. Yao, X. and Marshall, A. W., Characterizing Turbulent Ceiling Jet Dynamics with Salt-Water Modeling, Fire Safety Science-Proceedings of the Eighth International Symposium, Eds. Gottuk, G. T. and Lattimer, B. Y., pp. 927–938.

Appendix

Flammability Properties Archibald Tewarson, FM Global, Norwood, Massachusetts, USA These tables have been supplied by Dr Archibald Tewarson and are added as an aid in pursuing calculations for materials. Dr Tewarson has been singularly responsible for developing techniques for measuring material properties related to their fire characteristics, and for systematically compiling databases of these properties. His contributions are unique for this fledgling field of fire science. He has been a pioneer in using oxygen depletion and combustion products to measure the energy release rates in the combustion of materials under fire conditions. He has generously released these tables for use in the text. The tables were taken from A. Tewarson, Flammability of Polymers, Chapter 11 in Plastics and the Environment, A. L. Andrady (editor), John Wiley and Sons, Inc. Hoboken, N.J. 2003.

Fundamentals of Fire Phenomena James G. Quintiere # 2006 John Wiley & Sons, Ltd ISBN: 0-470-09113-4

410

APPENDIX

Table 1

Glass transition temperature (Tgl ), melting temperature (Tm ) and melt flow index (MFI) of polymers

Polymer

Polyethylene low density, PE-LD PE-high density, PE-HD Polypropylene, PP (atactic) Polyvinylacetate, PVAC Polyethyleneterephthalate, PET Polyvinyl chloride, PVC Polyvinylalcohol, PVAL PP (isotactic) Polystyrene, PS Polymethylmethacrylate, PMMA

Tgl ( C) Ordinary polymers 125

Tm ( C) 105–110 130–135 160–165 103–106 250

20 28 69 81 85 100 100 100–120 130, 160 High-temperature polymers Polyetherketone, PEK 119–225 Polyetheretherketone, PEEK 340 Polyethersulfone, PES 190 Halogenated polymers Perfluoro-alkoxyalkane, PFA 75 300–310 TFE,HFP,VDF fluoropolymer 200 115–125 TFE,HFP,VDF fluoropolymer 400 150–160 TFE,HFP,VDF fluoropolymer 500 165–180 Polyvinylidenefluoride, PVDF 160–170 Ethylenechlorotrifluoroethylene, ECTFE 240 Ethylenetetrafluoroethylene, ETFE 245–267 Perfluoroethylene-propylene, FEP 260–270 MFA 280–290 Tetrafluoroethylene, TFE 130 327

MFI (g/10 min)

1.4 2.2 21.5

9.0 2.1, 6.2

20 10 10

411

FLAMMABILITY PROPERTIES

Table 2 Char yield, vaporization/decomposition temperatures (Tv =Td ), limiting oxygen index (LOI) and UL 94 ratings for polymers Polymer

Tv =Td ( C) Char yield (%)

Ordinary polymers Poly(-methylstyrene) 341 0 Polyoxymethylene (POM) 361 0 Polystyrene (PS) 364 0 Poly(methylmethacrylate) (PMMA) 398 2 Polyurethane elastomer (PU) 422 3 Polydimethylsiloxane (PDMS) 444 0 Poly(acrylonitrile-butadiene-styrene) (ABS) 444 0 Polyethyleneterephthalate (PET) 474 13 Polyphthalamide 488 3 Polyamide 6 (PA6)-Nylon 497 1 Polyethylene (PE) 505 0 High-temperature polymers Cyanate ester of bisphenol-A (BCE) 470 33 Phenolic Triazine Cyanate Ester (PT) 480 62 Polyethylenenaphthalate (PEN) 495 24 Polysulfone (PSF) 537 30 Polycarbonate (PC) 546 25 Liquid crystal polyester 564 38 Polypromellitimide (PI) 567 70 Polyetherimide (PEI) 575 52 Polyphenylenesulfide (PPS) 578 45 Polypara(benzoyl)phenylene 602 66 Polyetheretherketone (PEEK) 606 50 Polyphenylsulfone (PPSF) 606 44 Polyetherketone (PEK) 614 56 Polyetherketoneketone (PEKK) 619 62 Polyamideimide (PAI) 628 55 Polyaramide (Kevlar1) 628 43 Polybenzimidazole (PBI) 630 70 Polyparaphenylene 652 75 Polybenzobisoxazole (PBO) 789 75 Halogenated polymers Poly(vinylchloride) (PVC) 270 11 Polyvinylidenefluoride (PVDF) 320–375 37 Poly(chlorotrifluoroethylene) (PCTFE) 380 0 Fluorinated Cyanate Ester 583 44 Poly(tetrafluoroethylene) (PTFE) 612 0 a b

Minimum oxygen concentration needed for burning at 20 C. V, vertical burning; HB, horizontal burning.

LOIa (%) UL 94 rankingb

18 15 18 17 17 30 18 21 (22) 21 18

HB HB HB HB HB HB HB HB HB HB HB

24 30 32 30 26 40 37 47 44 41 35 38 40 40 45 38 42 55 56

V-1 V-0 V-2 V-1 V-2 V-0 V-0 V-0 V-0 V-0 V-0 V-0 V-0 V-0 V-0 V-0 V-0 V-0 V-0

50 43–65 95 40 95

V-0 V-0 V-0 V-0 V-0

412

APPENDIX Table 3 Surface re-radiation loss q_ 00rr and heat of gasification (L) of polymers L (kJ/g)

Polymer/sample

00

q_ rr (kW/m2)

Ordinary polymers Filter paper 10 Corrugated paper 10 Douglas fir wood 10 Plywood/fire retarded (FR) 10 Polypropylene, PP 15 Polyethylene, PE, low density 15 PE-high density 15 Polyoxymethylene, POM 13 Polymethylmethacrylate, PMMA 11 Nylon 6,6 15 Polyisoprene 10 Acrylonitrile-butadiene-styrene, ABS 10 Styrene-butadiene 10 Polystyrene, PS foams 10–13 PS granular 13 Polyurethane, PU, foams-flexible 16–19 PU foams-rigid 14–22 Polyisocyanurate, PIU, foams 14–37 Polyesters/glass fiber 10–15 PE foams 12 High-temperature polymers Polycarbonate, PC 11 Phenolic foam 20 Phenolic foam/FR 20 Phenolic/glass fibers 20 Phenolic-aromatic polyamide 15 Halogenated polymers PE/25 % chlorine (Cl) 12 PE/36 % Cl 12 PE/48 % Cl 10 Polyvinylchloride, PVC, rigid 15 PVC plasticized 10 Ethylene-tetrafluoroethylene, ETFE 27 Perfluoroethylene-propylene, FEP 38 Ethylene-tetrafluoroethylene, ETFE 48 Perfluoro-alkoxyalkane, PFA 37

DSC

2.0 1.9 2.2 2.4 1.6

1.8 1.4

ASTM E 2058

3.6 2.2 1.8 1.0 2.0 1.8 2.3 2.4 1.6 2.4 2.0 3.2 2.7 1.3–1.9 1.7 1.2–2.7 1.2-5.3 1.2–6.4 1.4–6.4 1.4–1.7 2.1 1.6 3.7 7.3 7.8 2.1 3.0 3.1 2.5 1.7 0.9 2.4 0.8–1.8 1.0

387 404 356 308 293 325 298 274 310 346 296 341 284 305 257 428 418 330 252 447 572 565 341 334

537 606 397 454 411

PSF PEEK PC-1 PC-2 PC-3

Tv =Td ( C)

PE-1 PE-2 PE-3 PP-1 PP-2 PP-3 PP-4 PP-5 PP-6 PP-7 PP-8 PP-9 PE-PP-1 PE-PP-2 PVC-1 Nylon 66 Nylon 12 PMMA POM EPDM-1 EPDM-2 EPDM-3 SMC-ester TPO-1

Polymer

20

30 30

20

10 10

10

15 15 10 15

15

CHF (kW/m )

2

c (J/g K)

Ordinary polymers 0.94 2.15 0.95 2.03 0.95 2.12 0.90 2.25 0.90 2.48 1.19 1.90 1.21 1.76 1.11 1.95 0.93 2.20 0.90 2.22 1.06 2.08 1.04 1.93 0.91 1.98 0.88 2.15 1.95 1.14 1.50 1.69 1.04 1.79 1.19 2.09 1.41 1.92 1.15 1.75 1.16 1.39 1.21 1.48 1.64 1.14 0.97 1.87

(g/cm )

3

k (W/m K)

0.672 0.771 0.582 0.694 0.700

0.779 0.773 0.863 0.620 0.685 0.939 0.911 0.858 0.640 0.678 0.712 0.789 0.553 0.630 0.745 1.213 0.579 0.819 0.855 0.777 0.762 0.898 0.832 0.774 0.778 0.140

(kcÞ0:5 (kW s1/2/m2 K)

0.30 0.31 0.37 0.19 0.21 0.39 0.39 0.34 443 0.20 443 0.23 374 0.23 443 0.31 0.17 0.21 374 0.25 0.58 0.18 374 0.27 374 0.27 0.30 0.36 0.45 497 0.37 0.33 Average Standard deviation High temperature advanced-engineered polymers 580 1.24 1.30 0.28 580 1.32 1.80 0.25 1.12 1.68 0.18 497 1.18 1.51 0.27 1.19 2.06 0.20

443

Tig ( C)

Table 4 Thermophysical and ignition properties of polymersa

357

469 550

483

274 250

215

288 323 277 333

454

TRPExp

347 452 219 301 274

286 297 290 179 187 286 253 218 186 221 197 253 146 180 177 495 230 254 198 332 421 489 267 243

331

376 432

397

290 303

264

271 287 252 334

327

TRPCal from Tv =Td b Tig c

450 413 546 546 575

401 401 337 369 401 348 348

248 252 262 240 238 339 432 276 334 284 396

PTFE FEP ETFE PCTFE ECTFE PVDF CPVC

PEU-1 PEU-2 PEU-3 ABS-PVC-1 ABS-PVC-2 PS-1 PS-2 EPDM-1 EPDM-2 TPO-2 TPO-3

Tv =Td ( C)

PC-4 PC-5 PC-6 PC-7 PEI

Polymer

20

19

50 50 25 30 38 40 40

30 30 25

20

497

Tig ( C)

1.20 1.18 1.20 1.20 1.27

(g/cm )

3

2.18 1.10 1.20 1.20 1.40

c (J/g K) k (W/m K)

0.088 0.092 0.026 0.052 0.146 0.161 0.145 0.266 0.361 0.486 0.604

0.738 0.803 0.593 0.646 0.503 0.536 0.545 0.624 0.112

0.759 0.497 0.550 0.550 0.625 0.640 0.094

(kcÞ0:5 (kW s1/2/m2 K)

0.22 0.19 580 0.21 580 0.21 540 0.22 Average Standard deviation Highly halogenated advanced-engineered polymers 700 2.18 1.00 0.25 700 2.15 1.20 0.25 540 1.70 0.90 0.23 580 2.11 0.90 0.22 613 1.69 1.00 0.15 643 1.70 1.30 0.13 643 1.50 0.90 0.22 Average Standard deviation Foams and expanded elastomers 0.11 1.77 0.04 0.09 1.56 0.06 0.02 1.65 0.02 487 0.10 1.35 0.02 0.07 2.02 0.15 0.09 1.70 0.17 497 0.13 1.62 0.10 0.44 2.30 0.07 0.41 1.51 0.21 0.97 1.87 0.13 0.93 1.96 0.20

CHF (kW/m )

2

Table 4 (Continued)

146

73

654 680 481 460 450 506 435

455 455 435

434

TRPExp

20 21 6 11 32 51 60 68 113 128 227

281 306 188 226 192 176 179

326 195 289 289 347

69

24

502 546 308 362 299 334 340

308 308 325

362

TRPCal from Tv =Td b Tig c

394 305 380

PET-1 PET-2 Nylon 6

374 497

20

374

10

10

1.20 1.00 1.60 Fabrics 0.66 0.69 0.12 1.32 1.56 2.19

1.37 1.05 1.24 0.09 0.04 0.24

0.14 0.23 0.10 0.280 0.207 0.251

0.480 0.491 0.445

a

pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ Critical heat flux for piloted ignition (CHF), piloted ignition temperature (Tig ), thermal response parameter ðTRP ¼ 4 ðkcÞðTig T1 ÞÞ; b T v ; T d values were used in the calculation of TRP values. c T ig values from CHF were used in the calculation of TRP values.

269 261 267

PVC-2 PVC-glass PVC-elastomers

154

174

263

105 59 90

119 118 110

120

99

162

403 403 120 218

High density polyethylene, PE-HD Polyethylene, PE Polypropylene, PP Polypropylene, PP PP/glass fibers (1082) Polystyrene, PS PS foam PS-fire retarded (FR) PS foam-FR Nylon Nylon 6 Nylon/glass fibers (1077) Polyoxymethylene, POM Polymethylmethacrylate, PMMA Polybutyleneterephthalate, PBT Polyethyleneterephthalate, PET Acrylonitrile-butadiene-styrene, ABS ABS-FR ABS-PVC Vinyl thermoplastic elastomer Polyurethane, PU foam Thermoplastic PU-FR EPDM/Styrene acrylonitrile,SAN Polyester/glass fibers (30 %) Isophthalic polyester Isophthalic polyester/glass fiber (77 %) Polyvinyl ester Polyvinyl ester/glass fiber (69 %) Polyvinyl ester/glass fiber (1031) 259 176 609 718 299 212 5198 NI 12 302 486 NI 256 480 332 646

417 NIb 244 NI 1923 700

20

Polymers

278

193

168

25

115 172 120 235

130

193

77

73

63

171

30

74 36 113 116 68 66 61 1271 1 60 68 309 59 91 78 104

97 28 90 40 65 115

91 159 35 86

40

109 38 77 38 78 74

43

74 53

24

18

47

27

58

50

Ignition time (seconds) 00 Incident radiant heat flux, q_ e (kW/m2)

39 39 60 1 38 36

24 11 59 42

31

51

50

41

47

70

34

21

23

75

18

13

13

100

364 526 291 556 377 556 168 667 221 333 379 359 357 222 588 435 317 556 357 294 76 500 417 256 296 426 263 444 429

(kW s1/2/m2)

TRPExp

Table 5 Ignition-time measured in the ASTM E 1354 cone calorimeter and thermal response parameter values derived from the data

Polyvinyl ester/glass fiber (1087) Epoxy Epoxy/glass fiber (69 %) Epoxy/glass fiber (1003) Epoxy/glass fiber (1006) Epoxy/glass fiber (1040) Epoxy/glass fiber (1066) Epoxy/glass fiber (1067) Epoxy/glass fiber (1070) Epoxy/glass fiber (1071) Epoxy/glass fiber (1089) Epoxy/glass fiber (1090) Epoxy/graphite fiber (1091) Epoxy/graphite fiber (1092) Epoxy/graphite fiber (1093) Cyanate ester/glass fiber (1046) Acrylic/glass fiber Kydex acrylic paneling, FR Polycarbonate, PC-1 PC-2 Cross linked polyethylene (XLPE) Polyphenylene oxide, PPO-polystyrene (PS) PPO/glass fibers Polyphenylenesulfide, PPS/glass fibers (1069) PPS/graphite fibers (1083) PPS/glass fibers (1084) PPS/graphite fibers (1085) Polyarylsulfone/graphite fibers (1081) Polyethersulfone/graphite fibers (1078) Polyetheretherketone, PEEK/glass fibers (30 %) PEEK/graphite fibers (1086) Polyetherketoneketone, PEKK/glass fibers (1079) Bismaleimide, BMI/graphite fibers (1095) Bismaleimide, BMI/graphite fibers (1096) Bismaleimide, BMI/graphite fibers (1097) NI

553 200 NI 6400 750 479 465

337 320

NI 237 503 NI

NI NI NI NI NI NI

140 209 229 128 535 479 NI 275 338 199

198 159

281

252

172 120

390

148 38 182 144 105 87 45

100 75

141

244 173 122 172 142 307 223

105

76 94 58 101

62 57 50 49 18 48 63 63 34 105 120

12 75 45 35 39 35

80 92 42 60 66

57 69 70 59 40 47

73 23 13 14 24 30 18 60 54 53 32 44 20

22

42 53 22 36 37

30 26 48 33 19 21

19 14 9 9 18 23 10 40 34 28 23 28 10

11

312 457 388 555 397 512 288 433 517 334 665 592 484 493 554 302 180 233 455 370 385 455 435 588 330 623 510 360 352 301 514 710 513 608 605

Bismaleimide, BMI/graphite fibers (1098) Phenolic/glass fibers (45 %) Phenolic/glass fibers (1099) Phenolic/glass fibers (1100) Phenolic/glass fibers (1101) Phenolic/glass fibers (1014) Phenolic/glass fibers (1015) Phenolic/glass fibers (1017) Phenolic/glass fibers (1018) Phenolic/graphite fibers (1102) Phenolic/graphite fibers (1103) Phenolic/graphite fibers (1104) Phenolic/PE fibers (1073) Phenolic/aramid fibers (1074) Polyimide/glass fibers (1105) Douglas fir Hemlock Wool Acrylic fiber Polyvinylchloride, PVC flexible-1 PVC flexible-2 PVC flexible-3 (LOI 25 %) PVC-FR flexible-1 PVC-FR flexible-2 PVC-FR (Sb2O3) flexible-4 (LOI 30 %) PVC-FR (triaryl phosphate) flexible-5 (LOI 34 %) PVC-FR (alkyl aryl phosphate) flexible-6 (LOI 28 %) PVC rigid-1

Polymers

254 307 24 52 117 102 119 236 176 136 278 114 5159

423

20

NI NI NI NI NI NI NI NI NI NI 714 1110 NI

NI

25

84 103 72

61

73 15 28

30

34 35 11 16 27 21 41 47 36 64 69 49 73

214

40

37 45 35

25

19 9 19

104 187 129 163 175

110 165 121 125 210 214 238 180 313

50

Ignition time (seconds) 00 Incident radiant heat flux, q_ e (kW/m2)

Table 5 (Continued)

45

12 14

11 15

12

70

55 73 113 83 140 79 34 88 28 33 75

33

32

75

22 40 25 54 59 43 88 45 20 65 10 15 55

27

100

TRPExp

515 683 409 728 382 738 765 641 998 684 398 982 267 278 844 222 175 232 175 244 333 285 222 263 397 345 401 385

(kW s1/2/m2)

NI: no ignition

PVC rigid-2 PVC rigid-3 PVC rigid-1 (LOI 50 %) PVC rigid-2 Chlorinated PVC (CPVC)

3591 5171 NI NI NI 487 320

85 187 276 153 621 82 87 372

48 43

417 357 388 390 1111

Ref.

44 42 44 42 43 42 42 44 43 42 42 42 44 42 42 42 42 42 42 44 44 44 44 44 43 43 44 44

High-density polyethylene, HDPE Polyethylene, PE Polypropylene, PP Polypropylene, PP PP/glass fibers (1082) Polystyrene, PS Nylon Nylon 6 Nylon/glass fibers (1077) Polyoxymethylene, POM Polymethylmethacrylate, PMMA Polybutyleneterephthalate, PBT Acrylonitrile-butadiene-styrene, ABS ABS ABS-FR ABS-PVC Vinyl thermoplastic elastomer Polyurethane, PU foam EPDM/Styrene acrylonitrile,SAN Polyester/glass fibers (30 %) Isophthalic polyester Isophthalic polyester/glass fibers (77 %) Polyvinyl ester Polyvinyl ester/glass fibers (69 %) Polyvinyl ester/glass fibers (1031) Polyvinyl ester/glass fibers (1087) Epoxy Epoxy/glass fibers (69 %)

25

30

Ordinary polymers 453 866 913 377 693 1170 187 723 517 593 802 67 290 409 850 683 947 614 224 224 19 290 737 NI 582 861 173 170 341 471 251 230 75 377 392 453 164 161

20

560 172

360 665 1313 994 944 402 291 77 710 956 167 985 205 534 253

1101 1313 863

944 1408 1095 1509

40

706 202

231 985 198 755 222 119

1147

1272 96

361

1304

1133

50

1311 419 409 120 1221 1215

566 988 1984

1555 2019

2421

2735

70

Incident radiant heat flux, q_ e (kW/m2)

00

00 Q_ ch (peak) (kW/m2)

Peak heat release rate measured in the ASTM E 1354 cone calorimeter

Polymers

Table 6

139 499

116

484

75

166 557

135

432

100

21 37 32 25 6 17 30 21 1 6 12 23 14 12 4 4 2 19 10 6 20 2 13 2 1 2 11 2

hc =L ()

43 43 43 43 43 43 43 43 43 43 43 43 43 42 42 42 42 42 42 43 43 43 43 43 43 44 43 43 43 43 43 43 44 43 43 43

Epoxy/glass fibers(1003) Epoxy/glass fibers(1006) Epoxy/glass fibers (1040) Epoxy/glass fibers (1066) Epoxy/glass fibers (1067) Epoxy/glass fibers (1070) Epoxy/glass fibers (1071) Epoxy/glass fibers (1089) Epoxy/glass fibers (1090) Epoxy/graphite fibers (1091) Epoxy/graphite fibers (1092) Epoxy/graphite fibers (1093) Cyanate ester/glass fibers (1046) Kydex acrylic paneling, FR

Polycarbonate, PC-1 PC-2 Cross linked polyethylene (XLPE) Polyphenylene oxide, PPO-polystyrene (PS) PPO/glass fibers Polyphenylenesulfide, PPS/glass fibers (1069) PPS/graphite fibers (1083) PPS/glass fibers (1084) PPS/graphite fibers (1085) Polyarylsulfone/graphite fibers (1081) Polyethersulfone/graphite fibers (1078) Polyetheretherketone, PEEK/glass fibers (30 %) PEEK/graphite fibers (1086) Polyetherketoneketone, PEKK/glass fibers (1079) Bismaleimide, BMI/graphite fibers (1095) Bismaleimide, BMI/graphite fibers (1096) Bismaleimide, BMI/graphite fibers (1097) Bismaleimide, BMI/graphite fibers (1098) Phenolic/glass fibers (45 %) Phenolic/glass fibers (1099) Phenolic/glass fibers (1100) Phenolic/glass fibers (1101) NI NI NI

NI 160 128 NI NI 423

117 High-temperature polymers 16 144 88 219 154 NI NI NI NI NI NI NI

231 230 175 20 39 118 NI 164 105 121

159 81

214

35

429 420 192 265 276

176

74 165 66 66 47

176

48 94 24 11 109 14 21

52

189 171 130

294 181 40 266 213 196 93 178 114

342 535 268 301 386

242

102 120 57

54 45 213 245 172 91

71 60 88 66 47 41

191 182 246 271 300 262 141 217 144 197 242 244 196

122 163 96

85 74 270 285 168 146

183 80 150 126 60 65

335 229 232 489 279 284 202 232 173 241 242 202 226

21 14 5 2 6 3 2 2 1 1 0.3 7 1 1 1 2 (1) 1 1 1 2 1

2 2 2 3 1 2 2 2 1 2 2 3 2 3

Ref. 43 43 43 43 43 43 43 43 43 44 43 42 44 44 44 44 44 44 42 42 42 44 44 42

Polymers

Phenolic/glass fibers (1014) Phenolic/glass fibers (1015) Phenolic/glass fibers (1017) Phenolic/glass fibers (1018) Phenolic/graphite fibers (1102) Phenolic/graphite fibers (1103) Phenolic/graphite fibers (1104) Phenolic/PE fibers (1073) Phenolic/aramid fibers (1074) Phenolic insulating foam Polyimide/glass fibers (1105)

Douglas fir Hemlock

Wool Acrylic fibers

PVC flexible-3 (LOI 25 %) PVC-FR (Sb2O3) flexible-4 (LOI 30 %) PVC-FR (triaryl phosphate) flexible-5 (LOI 34 %) PVC rigid-1 PVC rigid-2 PVC rigid-3 PVC rigid-1 (LOI 50 %) PVC rigid-2 Chlorinated PVC (CPVC)

NI Wood

NI NI NI NI NI NI NI NI NI

25

218

30

Textiles 212 261 300 358 Halogenated polymers 126 148 89 137 96 150 40 75 102 NI 90 NI 101 25

237 233

20

Table 6 (Continued)

240 189 185 175 111 183 107 137 84

307 346

221 236

17

40

155 157

250 185 176

286 343

243

177 71 98 51 19 40

81 82 190 132

50

93

191 126 190

196

70

Incident radiant heat flux, q_ e (kW/m2)

00

00 Q_ ch (peak) (kW/m2)

97 76 115 56 159 183 87 141 93 29 78

75

85

133 80 141 68 196 189 101 234 104

100

5 5 5 3 2 2 3 3 1

5 6

(-) (-)

1 (1) 1 1 2 0.2 1 3 1 1 1

hc =L ()

FLAMMABILITY PROPERTIES

423

Table 7 Average yields of products and heat of combustion for polymers from parts of a minivan from the data measured in the ASTM E 2058 fire propagation apparatusa yj (g/g) Part description

Fabric, exposed Cover Shelf, main panel Structure Intake tube Backing Large ducts Inner boot Cotton shoddy Reservoir Structure High-density foam Face Fuel tank shield Cover Seal, foam Tank Lens Cover Grommet, wire harness cap a b c

Polymers

CO

CO2

Head liner Nylon 6 0.086 2.09 Instrument panel PVC 0.057 1.72 PC 0.051 1.86 Resonator PP 0.041 2.46 EPDM 0.045 2.51 Kick panel insulation PVC 0.061 1.26 Air ducts PP 0.056 2.52 Steering column boot NR 0.061 1.87 Cott/polyesterc 0.039 2.17 Brake fluid reservoir PP 0.058 2.41 Windshield wiper tray SMC 0.061 1.86 Sound reduction fender insulation PS 0.064 1.8 Hood liner insulation PET 0.041 1.47 Wheel well cover PP 0.054 2.45 HVAC unit PP 0.057 2.49 ABS-PVC 0.089 1.62 Fuel tank PE 0.032 2.33 Headlight PC 0.049 1.67 Battery casing PP 0.045 2.59 Bulkhead insulation engine side HDPF 0.064 2.67 2

hc HCb

Smoke

(kJ/g)

0.001

0.045

28.8

0.005 0.002

0.109 0.105

24.4 20.2

0.002 0.001

0.072 0.100

34.6 33.8

0.006

0.070

17.4

0.004

0.080

35.5

0.003 0.002

0.130 0.087

25.6 29.4

0.011

0.072

33.9

0.003

0.100

25.5

0.002

0.098

24.6

0.003

0.022

20

0.002

0.065

34.5

0.002 0.001

0.060 0.060

35 22.6

0.005

0.042

32.7

0.004

0.113

18.2

0.004

0.071

36.2

0.012

0.058

38.2

Combustion in normal air at 50 kW/m of external heat flux in the ASTM E 2058 apparatus. HC-total hydrocarbons. Cotton/polyester: mixture of cotton, polyester, and other fibers.

424

APPENDIX

Table 8 Average heat of combustion and yields of products for polymers from the data measured in the ASTM E 2058 fire propagation apparatus yj (g/g) Polymer Polyethylene, PE Polypropylene, PP Polystyrene, PS Polystyrene foam Wood Polyoxymethylene, POM Polymethylmethacrylate, PMMA Polyester Nylon Flexible polyurethane foams Rigid polyurethane foams Polyetheretherketone, PEEK Polysulfone, PSO Polyethersulfone, PES Polyetherimide, PEI Polycarbonate, PC

Composition Ordinary-polymers CH2 CH2 CH CH1:1 CH1:7 O0:73 CH2:0 O CH1:6 O0:40 CH1:4 O0:22 CH1.8O0.17N0.17 CH1.8O0.32N0.06 CH1:1 O0.21N0.10 High temperature polymers CH0.63O0.16 CH0.81O0.15S0.04

CH0.65O0.16N0.05 CH0.88O0.19 Halogenated polymers PE þ 25 % Cl CH1.9Cl0.13 PE þ 36 % Cl CH1.8Cl0.22 PE þ 48 % Cl CH1.7Cl0.36 Polyvinylchloride, PVC CH1.5Cl0.50 Chlorinated PVC CH1.3Cl0.70 Polyvinylidenefluoride, PVDF CHF Polyethylenetetrifluoroethylene, ETFE CHF Polyethylenechlorotrifluoroethylene, ECTFE CHCl0.25F0.75 Polytetrafluoroethylene, TFE CF2 Polyfluoroalkoxy, PFA CF1.6 Polyfluorinated ethylene propylene, FEP CF1.8 a

HC-total gaseous hydrocarbon.

CO

CO2 HCa

hc Smoke (kJ/g)

0.024 0.024 0.060 0.061 0.004 0.001 0.010 0.075 0.038 0.028 0.036

2.76 2.79 2.33 2.32 1.30 1.40 2.12 1.61 2.06 1.53 1.43

0.007 0.006 0.014 0.015 0.001 0.001 0.001 0.025 0.016 0.004 0.003

0.060 0.059 0.164 0.194 0.015 0.001 0.022 0.188 0.075 0.070 0.118

38.4 38.6 27.0 25.5 12.6 14.4 24.2 20.1 27.1 17.6 16.4

0.029 0.034 0.040 0.026 0.054

1.60 1.80 1.50 2.00 1.50

0.001 0.001 0.001 0.001 0.001

0.008 0.020 0.021 0.014 0.112

17.0 20.0 16.7 20.7 16.7

0.042 0.051 0.049 0.063 0.052 0.055 0.035 0.095 0.092 0.099 0.116

1.71 0.83 0.59 0.46 0.48 0.53 0.78 0.41 0.38 0.42 0.25

0.016 0.017 0.015 0.023 0.001 0.001 0.001 0.001 0.001 0.001 0.001

0.115 0.139 0.134 0.172 0.043 0.037 0.028 0.038 0.003 0.002 0.003

22.6 10.6 5.7 7.7 6.0 5.4 7.3 4.5 2.8 1.8 1.0

FLAMMABILITY PROPERTIES

425

Table 9 Average effective (chemical) heat of combustion and smoke yield calculated from the data measured in the ASTM E 1354 cone calorimetera Polymers Ordinary polymers High-density polyethylene, HDPE Polyethylene, PE Polypropylene, PP Polypropylene, PP PP/glass fibers (1082) Polystyrene, PS PS-FR PS foam PS foam-FR Nylon Nylon 6 Nylon/glass fibers (1077) Polyoxymethylene, POM Polymethylmethacrylate, PMMA Polybutyleneterephthalate, PBT Polyethyleneterephthalate, PET Acrylonitrile-butadiene-styrene, ABS ABS ABS-FR ABS-PVC Vinyl thermoplastic elastomer Polyurethane, PU foam Thermoplastic PU-FR EPDM/styrene acrylonitrile,SAN Polyester/glass fibers (30 %) Isophthalic polyester Isophthalic polyester/glass fibers (77 %) Polyvinyl ester Polyvinyl ester/glass fibers (69 %) Polyvinyl ester/glass fibers (1031) Polyvinyl ester/glass fibers (1087) Epoxy Epoxy/glass fibers (69 %) Epoxy/glass fibers (1003) Epoxy/glass fibers (1006) Epoxy/glass fibers (1040) Epoxy/glass fibers (1066) Epoxy/glass fibers (1067) Epoxy/glass fibers (1070) Epoxy/glass fibers (1071) Epoxy/glass fibers (1089) Epoxy/glass fibers (1090) Epoxy/graphite fibers (1091) Epoxy/graphite fibers (1092) Cyanate ester/glass fibers (1046) Acrylic/glass fibers Kydex acrylic paneling, FR

hc (MJ/kg)

ysmoke (g/g)

40.0 43.4 44.0 42.6 NR 35.8 13.8 27.7 26.7 27.9 28.8 NR 13.4 24.2 20.9 14.3 30.0 29.4 11.7 17.6 6.4 18.4 19.6 29.0 16.0 23.3 27.0 22.0 26.0 NR NR 25.0 27.5 NR NR NR NR NR NR NR NR NR NR NR NR 17.5 10.2

0.035 0.027 0.046 0.043 0.105 0.085 0.144 0.128 0.136 0.025 0.011 0.089 0.002 0.010 0.066 0.050 0.105 0.066 0.132 0.124 0.056 0.054 0.068 0.116 0.049 0.080 0.032 0.076 0.079 0.164 0.128 0.106 0.056 0.142 0.207 0.058 0.113 0.115 0.143 0.149 0.058 0.086 0.082 0.049 0.103 0.016 0.095

426

APPENDIX

Table 9 (Continued) Polymers

hc (MJ/kg)

High-temperature polymers and composites Polycarbonate, PC-1 21.9 PC-2 22.6 Cross linked polyethylene (XLPE) 23.8 Polyphenylene oxide, PPO-polystyrene (PS) 23.1 PPO/glass fibers 25.4 Polyphenylenesulfide, PPS/glass fibers (1069) NR PPS/graphite fibers (1083) NR PPS/glass fibers (1084) NR PPS/graphite fibers (1085) NR Polyarylsulfone/graphite fibers (1081) NR Polyethersulfone/graphite fibers (1078) NR Polyetheretherketone, PEEK/glass fibers (30 %) 20.5 PEEK/graphite fibers (1086) NR Polyetherketoneketone, PEKK/glass fibers (1079) NR Bismaleimide, BMI/graphite fibers (1095) NR Bismaleimide, BMI/graphite fibers (1096) NR Bismaleimide, BMI/graphite fibers (1097) NR Bismaleimide, BMI/graphite fibers (1098) NR Phenolic/glass fibers (45 %) 22.0 Phenolic/glass fibers (1099) NR Phenolic/glass fibers (1100) NR Phenolic/glass fibers (1101) NR Phenolic/glass fibers (1014) NR Phenolic/glass fibers (1015) NR Phenolic/glass fibers (1017) NR Phenolic/glass fibers (1018) NR Phenolic/graphite fibers (1102) NR Phenolic/graphite fibers (1103) NR Phenolic/graphite fibers (1104) NR Phenolic/PE fibers (1073) NR Phenolic/aramid fibers (1074) NR Phenolic insulating foam 10.0 Polyimide/glass fibers (1105) NR Wood Douglas fir 14.7 Hemlock 13.3 Textiles Wool 19.5 Acrylic fiber 27.5 Halogenated polymers PVC flexible-3 (LOI 25 %) 11.3 PVC-FR (Sb2O3) flexible-4 (LOI 30 %) 10.3 PVC-FR (triaryl phosphate) flexible-5 (LOI 34 %) 10.8 PVC rigid-1 8.9 PVC rigid-2 10.8 PVC rigid-3 12.7 PVC rigid-1 (LOI 50 %) 7.7 PVC rigid-2 8.3 Chlorinated PVC (CPVC) 5.8

ysmoke (g/g) 0.098 0.087 0.026 0.162 0.133 0.063 0.075 0.075 0.058 0.019 0.014 0.042 0.025 0.058 0.077 0.096 0.095 0.033 0.026 0.008 0.037 0.032 0.031 0.031 0.015 0.009 0.039 0.041 0.021 0.054 0.024 0.026 0.014 0.010 0.015 0.017 0.038 0.099 0.078 0.098 0.103 0.112 0.103 0.098 0.076 0.003

CH2 CH2 CH CH1:1 CH1:7 O0:73 CH2:0 O CH1:6 O0:40 CH1:4 O0:22 CH2:0 O0:50 CH0:80 O0:40 CH1:1 N0:07 CH1:8 O0:17 N0:17 CH1:8 O0:32 N0:06 CH1:1 O0:21 N0:10

Polyethylene (PE) Polypropylene (PP) Polystyrene (PS) Polystyrene foam Wood Polyoxymethylene (POM) Polymethylmethacrylate (PMMA) Polyester Polyvinylalcohol Polyethyleneterephthalate (PET) Polyacrylonitrile-butadiene-styrene (ABS) Nylon Flexible polyurethane foam Rigid polyurethane foam

HC: hydrocarbons.

Composition

s (g/g) 14.7 14.7 13.2 13.4 5.8 4.6 8.2 10.1 7.8 7.2 13.2 11.2 9.4 9.8

M (g/mole) 14.0 14.0 13.0 13.1 25.4 30.0 20.0 16.9 22.0 19.2 14.1 18.9 19.7 17.8

43.6 43.4 39.2 39.2 16.9 15.4 25.2 32.5 21.3 23.2 38.1 30.8 25.3 25.9

hc;T (kJ/g) 3.14 3.14 3.38 3.36 1.73 1.47 2.20 2.60 2.00 2.29 3.13 2.33 2.24 2.47

CO2

2.00 2.00 2.15 2.14 1.11 0.93 1.40 1.65 1.27 1.46 1.99 1.48 1.42 1.57

CO

yi; max

1.14 1.14 1.23 1.22 0.63 0.53 0.80 0.95 0.73 0.83 1.14 0.85 0.81 0.90

HC

Composition, molecular weight, stoichiometric air-to-fuel ratio (s), net heat of complete combustion (hc;T ), and maximum possible stoichiometric yields of major products (yi;max ) for ordinary polymers

Polymers

Table 10

0.86 0.86 0.92 0.92 0.48 0.40 0.60 0.71 0.55 0.63 0.85 0.63 0.61 0.67

Smoke

Composition CH0.60O0.15 CH0.62O0.15 CH0.63O0.16 CH0.53O0.20 CH0.88O0.19 CH0.71O0.29 CHO0.13 CHO0.33 CH0.62O0.08 CHO0.14 CH0.60N0.20 CH0.67S0.17 CH0.43O0.14N0.14 CH0.45O0.23N0.09 CH0.65O0.16N0.05 CH0.71O0.14N0.14 CH0.71O0.14N0.14 CH0.71O0.14N0.14 CH0.81O0.15S0.04 CH0.67O0.25S0.08

Polyetherketoneketone (PEKK) Polyetherketone (PEK) Polyetheretherketone (PEEK) Polyamideimide (PAI) Polycarbonate (PC) Polyethylenenaphthalate (PEN) Polyphenyleneoxide (PPO) Polybutanediolterephthalate (PBT) Polybenzoylphenylene Phenol-formaldehyzde Polybenzimidazole (PBI) Polyphenylenesulfide (PPS) Polyphenylenebenzobisoxazole (PBO) Polyimide (PI) Polyetherimide (PEI) Polyaramide Polyphenylenebenzobisoxazole Aramid-arylester copolymer Polysulfone (PSF) Polyphenyleneethersulfone PES

15.0 15.0 15.2 15.7 15.9 17.4 15.0 18.3 13.9 15.2 15.4 18.0 16.6 17.4 16.0 16.9 16.9 16.9 16.4 19.3

(g/mole)

M

9.8 9.9 9.7 9.0 9.7 8.2 10.9 8.2 11.1 10.6 12.0 10.2 9.7 8.6 9.8 10.1 10.1 10.1 9.8 8.0

(g/g)

s

24.4 29.4 24.7

25.5 28.4

30.8 28.2

30.3 30.2 30.4 24.2 30.1

(kJ/g)

hc;T

2.93 2.93 2.90 2.80 2.76 2.54 2.93 2.41 3.18 2.89 2.86 2.44 2.65 2.53 2.76 2.60 2.60 2.60 2.68 2.28

CO2 1.87 1.86 1.84 1.78 1.76 1.61 1.87 1.53 2.02 1.84 1.82 1.55 1.68 1.61 1.75 1.66 1.66 1.66 1.71 1.45

CO 1.07 1.07 1.05 1.02 1.01 0.92 1.07 0.88 1.16 1.05 1.04 0.89 0.96 0.92 1.00 0.95 0.95 0.95 0.98 0.83

HC 0.80 0.80 0.79 0.76 0.75 0.69 0.80 0.66 0.87 0.79 0.78 0.67 0.72 0.69 0.75 0.71 0.71 0.71 0.73 0.62

Smoke

yi;max (g/g)

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.35 0.00 0.23 0.14 0.09 0.22 0.22 0.22 0.00 0.00

HCN

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.60 0.00 0.39 0.24 0.16 0.38 0.38 0.38 0.00 0.00

NO2

Composition, molecular weight, stoichiometric air-to-fuel ratio (s), net heat of complete combustion (hc;T ), and maximum possible stoichiometric yields of major products (yi;max ) for high temperature polymers

Polymers

Table 11

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.59 0.00 0.00 0.00 0.00 0.00 0.00 0.14 0.27

SO2

Polyethylenechlorotrifluoroethylene, ECTFE Polychlorotrifluoroethylene, CTFE

Polytetrafluoroethylene, TFE Polyperfluoroalkoxy, PFA Polyfluorinatedethylenepropylene,FEP

Polyvinylfluoride Polyvinylidenefluoride (PVDF) Polyethylenetrifluoroethylene, ETFE

PE þ 25 % Cl PE þ 36 % Cl Polychloropropene PE þ 48 % Cl Polyvinylchloride, PVC Chlorinated PVC Polyvinylidenechloride, PVCl2

Polymers

M (g/mole) s (g/g)

(kJ/g)

Carbon-hydrogen-chlorine atoms 18.5 10.7 31.6 CH1.9Cl0.13 21.5 8.9 26.3 CH1.8Cl0.22 23.8 7.2 25.3 CH1.3Cl0.30 26.3 7.0 20.6 CH1.7Cl0.36 31.0 5.5 16.4 CH1.5Cl0.50 37.8 4.2 13.3 CH1.3Cl0.70 CHCl 48.0 2.9 9.6 Carbon-hydrogen-fluorine atoms 23.0 7.5 13.5 CH1.5F0.50 CHF 32.0 4.3 13.3 CHF 32.0 4.3 12.6 Carbon-fluorine atoms 50.0 2.7 6.2 CF2 42.6 3.2 5.0 CF1.6 46.2 3.0 4.8 CF1.8 Carbon-hydrogen-chlorine-fluorine atoms 36.0 3.8 12.0 CHCl0.25F0.75 58.0 2.4 6.5 CCl0.50F1.5

Composition

hc;T

1.52 1.30 1.18 1.06 0.90 0.74 0.58 1.22 0.88 0.88 0.56 0.66 0.61 0.78 0.48

1.91 1.38 1.38 0.88 1.03 0.95 1.22 0.76

CO

2.38 2.05 1.85 1.67 1.42 1.16 0.92

CO2

0.44 0.28

0.32 0.38 0.35

0.70 0.50 0.50

0.87 0.74 0.56 0.61 0.52 0.42 0.33

HC

0.33 0.21

0.24 0.28 0.26

0.52 0.38 0.38

0.65 0.56 0.50 0.46 0.39 0.32 0.25

Smoke

yi; max (g/g)

0.25 0

0.00 0.00 0.00

0.00 0.00 0.00

0.25 0.37 0.45 0.49 0.58 0.67 0.75

HCl

Table 12 Composition, molecular weight, stoichiometric air-to-fuel ratio (s), net heat of complete combustion (hc;T ), and maximum possible stoichiometric yields of major products (yi;max ) for halogenated polymers

0.42 0

0.00 0.00 0.00

0.43 0.63 0.63

0.00 0.00 0.00 0.00 0.00 0.00 0.00

HF

430

APPENDIX

Table 13 Composition, molecular weight and combustion properties of hydrocarbons and alcohols Fluid

M Tig; auto Composition (g/mole) ( C)

Gasoline Hexane Heptane Octane Nonane Decane Undecane Dodecane Tridecane Kerosine Hexadecane Mineral oil Motor oil Corn Oil Benzene Toluene Xylene Methanol Ethanol Propanol Butanol na: not available.

nab C6H14 C7H16 C8H18 C9H20 C10H22 C11H24 C12H26 C13H28 C14H30 C16H34 na na na C6H6 C7H8 C8H10 CH4O C2H6O C3H8O C4H10O

na 86 100 114 128 142 156 170 184 198 226 466 na na 78 92 106 32 46 60 74

371 225 204 206 205 201 na 203 na 260 202 na na 393 498 480 528 385 363 432 343

Tboil ( C)

hc (kJ/g)

L (kJ/g)

hc =L (kJ/kJ)

yco (g/g)

ysm (g/g)

33 69 98 125 151 174 196 216

41.0 41.5 41.2 41.0 40.8 40.7 40.5 40.4 40.3 40.3 40.1 na 29.3 22.3 27.6 27.7 27.8 19.1 25.6 29.0 31.2

0.482 0.500 0.549 0.603 0.638 0.690 0.736 0.777 0.806 0.857 0.911

85 83 75 68 64 59 55 52 50 47 44 72 62 54 75 82 67 19 33 46 58

0.010 0.009 0.010 0.010 0.011 0.011 0.011 0.012 0.012 0.012 0.012 na na na 0.067 0.066 0.065 0.001 0.001 0.003 0.004

0.038 0.035 0.037 0.038 0.039 0.040 0.040 0.041 0.041 0.042 0.042 na na na 0.181 0.178 0.177 0.001 0.008 0.015 0.019

232 287 360 na na 80 110 139 64 78 97 117

0.473 0.413 0.368 0.338 0.415 1.005 0.776 0.630 0.538

431

FLAMMABILITY PROPERTIES Table 14 Asymptotic flame heat flux values for the combustion of liquids and polymers 00

q_ f (kW/m2)

Fuels Physical state

Liquids

Polymers

Liquids Polymers Liquids Polymers

Liquids

Liquid Polymers

Polymers

Name

O2 > 30 % (small scale)

Aliphatic carbon-hydrogen atom containing fuels Heptane 32 Hexane Octane Dodecane Kerosine Gasoline JP-4 JP-5 Transformer fluids 23–25 Polyethylene, PE 61 Polypropylene, PP 67 Aromatic carbon-hydrogen atom containing fuels Benzene Toluene Xylene Polystyrene, PS 75 Aliphatic carbon-hydrogen -oxygen atom containing fuels Methanol 22 Ethanol Acetone Polyoxymethylene, POM 50 Polymethylmethacrylate, PMMA 57 Aliphatic carbon-hydrogen -nitrogen atom containing fuels Adiponitrile Acetonitrile Aliphatic carbon-hydrogen-oxygen-nitrogen atom containing fuels Toluene diisocyanate Polyurethane foams (flexible) 64–76 Polyurethane foams (rigid) 49–53 Aliphatic carbon-hydrogen-halogen atom containing fuels Polyvinylchloride, PVC 50 Ethylenetetrafluoroethylene, ETFE 50 Perfluoroethylene-propylene, FEP 52

Normal air (large scale)

41 40 38 28 29 35 34 38 22–25

44 34 37 71 27 30 24 60 34 35 28

432

APPENDIX Table 15

Limited oxygen index (LOI)a values at 20 C for polymers

Polymers Ordinary polymers Polyoxymethylene Cotton Cellulose acetate Natural rubber foam Polypropylene Polymethylmethacrylate Polyurethane foam Polyethylene Polystyrene Polyacrylonitrile ABS Poly(-methylstyrene) Filter paper Rayon Polyisoprene Epoxy Polyethyleneterephthalate (PET) Nylon 6 Polyester fabric Plywood Silicone rubber (RTV, etc.) Wool Nylon 6,6 Neoprene rubber Silicone grease Polyethylenephthalate (PEN) High-temperature polymers Polycarbonate Nomex1 Polydimethylsiloxane (PDMS) Polysulfone Polyvinyl ester/glass fibers (1031) Polyetheretherketone (PEEK) Polyimide (Kapton1) Polypromellitimide (PI) Polyaramide (kevlar1) Polyphenylsulfone (PPSF) Polyetherketone (PEK) a

LOI 15 16 17 17 17 17 17 18 18 18 18 18 18 19 19 20 21 21 21 23 23 24 24-29 26 26 32 26 29 30 31 34 35 37 37 38 38 40

Oxygen concentration needed to support burning.

Polymers Polyetherketoneketone (PEKK) Polypara(benzoyl)phenylene Polybenzimidazole (PBI) Polyphenylenesulfide (PPS) Polyamideimide (PAI) Polyetherimide (PEI) Polyparaphenylene Polybenzobisoxazole (PBO) Composites Polyethylene/Al2O3(50 %) ABS/glass fibers (20 %) Epoxy/glass fibers (65 %) Epoxy/glass fibers (65 %)-300 C Epoxy/graphite fibers (1092) Polyester/glass fibers (70 %) Polyester/glass fibers(70 %)-300 C Phenolic/glass fibers (80 %) Phenolic/glass fibers(80 %)-100 C Phenolic/Kevlarr (80 %) Phenolic/Kevlarr (80 %)-300 C PPS/glass fibers (1069) PEEK/glass fibers (1086) PAS/graphite (1081) BMI/graphite fibers (1097) BMI/graphite fibers (1098) BMI/glass fibers (1097) Halogenated polymers Fluorinated cyanate ester Neoprene Fluorosilicone grease Fluorocarbon rubber Polyvinylidenefluoride PVC (rigid) PVC (chlorinated) Polyvinylidenechloride (Saranr) Chlorotrifluoroethylene lubricants Fluorocarbon (FEP/PFA) tubing Polytetrafluoroethylene Polytrichlorofluorethylene

LOI 40 41 42 44 45 47 55 56 20 22 38 16 33 20 28 53 98 28 26 64 58 66 55 60 65 40 40 31–68 41–61 43–65 50 45–60 60 67–75 77–100 95 95

FLAMMABILITY PROPERTIES

433

Table 16 Smoke yield for various gases, liquids and solid fuels State

Generic name

Generic nature

Gas

Methane to butane Ethylene, propylene Acetylene, butadiene Alcohols, ketones Hydrocarbons

Aliphatic-saturated Aliphatic-unsaturated Aliphatic-highly unsaturated Aliphatic Aliphatic Aromatic Aliphatic (mostly) Aliphatic-oxygenated Aliphatic-highly fluorinated Aliphatic-unsaturated Aliphatic unsaturated-chlorinated Aromatic

Liquid Solid

Cellulosics Synthetic Polymers

Smoke yield (g/g) 0.013–0.029 0.043–0.070 0.096–0.125 0.008–0.018 0.037–0.078 0.177–0.181 0.008–0.015 0.001–0.022 0.002–0.042 0.060–0.075 0.078–0.099 0.131–0.191

Index

absorption coefficient, 15 activation energy definition, 78 typical values, 79 adiabatic furnace, 128 America Burning, xiv, 1, 9 autoignition calculation of, 105–106 temperatures, 84 theory of, 80–85 wood data, 182–184 Avrogadro’s number, 22 B number, Spalding, 8, 242 Biot number, 121, 172 BLEVE, 140–141 blocking factor, 248 boundary layer thickness, 236 Boussinesq assumption, 300, 302 Buckingham pi method, 378–379 burning velocity, See flame speed burning rate alternative control volume theory, 269–277 asymptotic values, 259 char effect, 230 chemical kinetic effects, 277–282 complex materials, 267 convective, 248–252 definition, 227 extinction behavior, 282–285 mass flux, 228 radiation effects, 255–259 stagnant layer theory, 233–242 water effects, 262–267 wood cribs, 268, 353, 362–364

Fundamentals of Fire Phenomena James G. Quintiere # 2006 John Wiley & Sons, Ltd ISBN: 0-470-09113-4

ceiling jet, 395, 406 Center for Fire Research (CFR), xiv, 9 chemical equilibrium, 19 chemical reaction complete, 20 stoichiometric, 21 Clausius-Clapeyron equation derivation, 141–143 use in burning rate, 240 compartment fires burning rate for, See burning rate developing fires, 358–361 fluid dynamics of, 342–347 fuel interaction, 351–355 fully developed, 340–341, 360–364 heat transfer for, 347–352 pool fires, 353, 364 smoke filling, 396 temperatures, 358–364, 398–399 Conseil International du Batiment (CIB), 362 conservation equations compartment fires, 356–357 control volume formulation, 54–67 dimensional analysis, 381–384 energy, 61–67 mass, 54 momentum, 59 plumes, 302–306, 312–313 species, 56–57 control volume description, 49 relationship to system, 54 Dalton’s law, 25, 139 Damkohler number in burning, 234

436

INDEX

with conduction, 121 critical values, 123 in flame spread, 210–211 uniform system, 83, 88 deflagration, 77, 88 detonation, 77, 88 Denmark Technical University (DTU), xv, 11 diffusion coefficient, 16 diffusion velocity, 16, 56 dimensionless groups, 377, 392–394 educational institutions, 11 emissivity in compartment fires, 349–351 definition, 15 enthalpy, 28 equivalence ratio, 21 Error function, 178–179 evaporation description, 137–140 dimensionless groups, 386–388 flash evaporation, 140 rates, 146–148 temperature distribution, 149 of water sprays, 400–401 evaporative cooling, 137 FM Global, See Factory Mutual Factory Mutual, xiv, 10 Fick’s law, 8, 16.56, 236 fire cost, 2 definition, 3–6 disasters, 3 history, 1–2 science of, 6, 8 fire plumes with combustion, 311–315 described, 299 eddy frequency, 298–300 entrainment fraction, 312 entrainment rate, 319–321 entrainment velocity, 300 Gaussian profile, 298 geometric effects, 312–313 intermittency, 301 line source results, 308–310 natural length scale, 307 point source results, 308–310 radiation fraction, 297, 302, 315–316 top hat profile, 300

transient effects, 326–329, 405–406 virtual origin, 317–318 Fire Research Station (FRS), xiv, 9 fireball, 141, 328–332 firepoint, 136 firepower compartments, 354 described, 8 dimensionless, 383 plumes, 302 First law of thermodynamics, 27, 34 flame definition, 3 flame heated length, 201, 368 flame height, See flame length flame length axisymmetric plume, 322–324 ceiling interaction, 325 in corner, 325 jet, 323–324 line plume, 325 wall fire, 207, 326 against wall, 326 flame propagation, See flame speed flame sheet description, 244 location, 247 temperature, 247 flame speed in houses, 217–219 on liquids, 216–217 porous media, 215–216 pre-mixed flames ideal, 88 limiting value, 93–95 measurement techniques, 89–90 theory, 90–93 solid surface effect of Damkohler number, 211–212 effect of orientation, 213–215 effect of velocity and oxygen, 212–213 exact solutions, 211 opposed flow, 192–193 thermally thick theory, 200–202 thermally thin theory, 194–196 transient effects, 198–200 wind-aided, 192–193 flame temperature adiabatic, 29, 93, 103–104 critical for extinction, 95, 103, 264, 277–278, 281 maximum in turbulent plumes, 316

INDEX flammability limits description, 98–102 diluent effect, 102 for fuel mixture, 104 lower, 95 pressure effect, 101 stoichiometric relations, 102–103 upper, 95 flashback, 90 flashover, 340–342, 365–367 flashpoint description, 98, 135 open, closed cup, 135 fluid mechanics review, 14–15 dimensionless flows, 386 Fourier’s law, 8, 15 Froude number defined, 377 plume invariant, 377 fuel-controlled, 341 gas mixtures, 23–25 Gibbs function, 141–142 Grashof number in burning, 249 definition, 16 in pressure modeling, 397 heat flux asymptotic for flames, 431 in burning, 228, 257 critical in flame spread, 198 critical in ignition, 174, 413–415 in fire, 166 for ignition, 171 measurement of, 170 in opposed flow flame spread, 202–203 in scaling inconsistencies, 389–392 surface re-radiation in burning, 412 for wall flames, 207–208 heat of combustion definition, 28 by heat of formations, 33 ideal, 36 liquids, 430 per oxygen, 36 polymers, 424–430 values in fire, 37, 228 vehicles components, 423

437

heat of formation definition, 30 values, 31 heat of gasification liquids, 150 modified for radiation, 256 modified for water, 263 polymers, 412 solids, 231–233 heat of vaporization, 135 heat release rate, (HRR), See firepower heat transfer in compartments, 347–351 conduction, 15 convection, 15, 167 definition, 8 dimensionless, 384–386 radiation, 15, 167, 169–170 review, 15–16 heat transfer coefficient compartment fires, 348 definition, 15 effective for radiation, 173 total, 173 humidity, relative, 143 ideal gas equation of state, 14, 24 ignition, See also autoignition, piloted ignition, spontaneous ignition glowing, 160, 183 liquids, 135 solids, 159 thermally thick theory, 176–180 thermally thin theory, 171–174 properties for common materials, 184–187 of wood, 181–184 internal energy, 26 jet flames, 308 Lewis number, 16, 239 limiting oxygen limit (LOI), 432 lower flammability limit, See flammability limit mass fraction, 23 mass transfer coefficient, 16 mixture fraction, 243, 301 mole, 21 mole fraction, 23 molecular weight, 21

438

INDEX

National Bureau of Standards (NBS) See NIST National Commission of Fire Prevention and Control, 1 National Fire Protection Association (NFPA), 8 National Institute of Standards and Technology (NIST) early program, xiv founding, 8 WTC investigation, 10–11 National Science Foundation Research for National Needs (NSF RANN), xiv, 9 Newton’s second law, 26, 59 partial pressure, 25 perfect gas, See ideal gas piloted ignition in liquids, 135–136 minimum energy for, 86 in solids, 160–161 theory, 85–86 Planck’s law, 15 Prandtl number, 16 pre-exponential coefficient, 79 pressure, 13 pressure modeling description, 392 burning rate, walls, 398 burning rate, tubes, 397 properties in burning, 253 extensive, 13, 26 flammability data, 409–433 ignition, 165 intensive, 13, 26 opposed flow flame spread, 204 pyrolysis, 20, 163–164 quenching diameter calculations, 108–109 definition, 95 theory, 95–97 radiation, See heat transfer reaction rate, 78–79 Reynolds number chemical, 235, 261 definition, 15, 377 Reynolds transport theorem, 50

saturation pressure, 139 saturation states, 139 saturation temperature definition, 135 function of pressure, 139 scale modeling described, 401 Froude modeling, 402–403 saltwater modeling, 403–406 Schmidt number, 16 self-heating, 118 Semenov diagram autoignition, 81, 106 burning rate, 279–281 compartment fires, 365–369 piloted ignition, 87, 107 propagation, 94, 108, 280 quenching diameter, 96, 109 shear stress, 14 Sherwood number, 16 Shvab-Zeldovich variable, 241 Society of Fire Protection Engineers, (SFPE), xv spontaneous ignition adiabatic ignition time, 129 approximate time for, 130 description, 117–119 experimental methods, 124–127 theory, 119–124 stoichiometric coefficient, 22 system, 25 temperature glass transition, 410 ignition, 137, 164, 413 melt, 410 thermodynamic definition, 13 vaporization, 242, 411 time scales burning, 198 chemical reaction, 85, 161–162 flow, 382 ignition, 161 thermal diffusion, 85 mixing, 162–163 pyrolysis, 163–164 spontaneous ignition, 127–130 thermal penetration depth in ignition, 176, 178 in flame spread, 201 thermochemistry, 19 thermodynamics, 13–14

INDEX Universal gas constant, 24 ventilation-limited, 21, 341 Weber number, 392–393 wood cribs, See burning rate work, 27 World Trade Center (WTC), 2, 10–11 yield definition, 35, 228 for liquids, 430 maximum possible, 427–429

for polymers, 424 smoke, 425–426, 433 for vehicle components, 423 zone model compartments, 355 described, 49 Zukoski number ceiling jet, 395 general, 383 heat losses, 385 mass number, 386 plume, 319

439